+ All Categories
Home > Documents > Solar Measurement Campaign in Vietnam

Solar Measurement Campaign in Vietnam

Date post: 18-Mar-2023
Category:
Upload: khangminh22
View: 0 times
Download: 0 times
Share this document with a friend
154
Solar Measurement Campaign in Vietnam Site Measurement Report 24 Months of Measurement Data Selection #: 1231900 Suntrace GmbH Grosse Elbstrasse 145c 22767 Hamburg www.suntrace.de The World Bank Global ESMAP Initiative Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized Public Disclosure Authorized
Transcript

Solar Measurement Campaign in Vietnam Site Measurement Report

24 Months of Measurement Data

Selection #: 1231900 Suntrace GmbH

Grosse Elbstrasse 145c

22767 Hamburg

www.suntrace.de

The World Bank

Global ESMAP Initiative

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

Pub

lic D

iscl

osur

e A

utho

rized

www.suntrace.de

Page 1 of 153

Disclaimer

Subject: Site Measurement Report – 24 months

Client: The World Bank

Global ESMAP initiative

Consultant: Suntrace GmbH Grosse Elbstrasse 145c 22767 Hamburg

E: [email protected] W: www.suntrace.de

Date of Report: 21 December 19

Authors: Simon Weber, Marko Schwandt

Reviewed by: Joana Zerbin

Dr. Richard Meyer

V1 19/12/2019 SW 20/12/2019 MA 21/12/2019 JZ Version Date Authored Checked Approved

www.suntrace.de

Page 2 of 153

Content Page

1 Introduction .......................................................................................................................... 6 2 SITE DESCRIPTION ............................................................................................................. 7

2.1 Overview of Stations ..................................................................................................................... 7

2.1.1 VNHAN ......................................................................................................................................... 9 2.1.2 VNDAN ....................................................................................................................................... 15

2.1.3 VNCEH ....................................................................................................................................... 21

2.1.4 VNSOB ....................................................................................................................................... 27 2.1.5 VNTRA ........................................................................................................................................ 34

2.1.6 Handover of the stations ............................................................................................................. 42 3 Quality Checks ....................................................................................................................... 44 4 Solar Measurement Results ..................................................................................................... 46

4.1 VNHAN ......................................................................................................................................... 46 4.1.1 Quality Checks for VNHAN .............................................................................................................. 46 4.1.2 Seasonal and Diurnal Characteristics for VNHAN ...................................................................... 47

4.1.3 Soiling Measurements for VNHAN ............................................................................................. 50 4.1.4 Temperature and Humidity for VNHAN ...................................................................................... 54 4.1.5 Wind Measurements for VNHAN ................................................................................................ 56

4.1.6 Measurement Statistics for VNHAN ............................................................................................ 57 4.2 VNDAN ....................................................................................................................................... 62 4.2.1 Quality Checks for VNDAN ......................................................................................................... 62

4.2.2 Seasonal and Diurnal Characteristics for VNDAN ...................................................................... 63

4.2.3 Soiling Measurements for VNDAN ............................................................................................. 66 4.2.4 Temperature and Humidity for VNDAN ...................................................................................... 68

4.2.5 Measurement Statistics for VNDAN ............................................................................................ 69

4.3 VNCEH ....................................................................................................................................... 74 4.3.1 Quality Checks for VNCEH ......................................................................................................... 74

4.3.2 Seasonal and Diurnal Characteristics for VNCEH ...................................................................... 75

4.3.3 Soiling Measurements for VNCEH ............................................................................................. 77 4.3.4 Temperature and Humidity for VNCEH ...................................................................................... 80

4.3.5 Wind Measurements for VNCEH ................................................................................................ 82

4.3.6 Measurement Statistics for VNCEH ............................................................................................ 82 4.4 VNSOB ....................................................................................................................................... 87

4.4.1 Quality Checks for VNSOB ......................................................................................................... 87 4.4.2 Seasonal and Diurnal Characteristics for VNSOB ...................................................................... 88

www.suntrace.de

Page 3 of 153

4.4.3 Soiling Measurements for VNSOB ............................................................................................. 90

4.4.4 Temperature, Humidity and Precipitation for VNSOB ................................................................. 92 4.4.5 Wind Measurements for VNSOB ................................................................................................ 94

4.4.6 Measurement Statistics for VNSOB ............................................................................................ 94

4.5 VNTRA ........................................................................................................................................ 99 4.5.1 Quality Checks for VNTRA ......................................................................................................... 99

4.5.2 Seasonal and Diurnal Characteristics for VNTRA .................................................................... 100

4.5.3 Soiling Measurements for VNTRA ............................................................................................ 102 4.5.4 Temperature and Humidity for VNTRA ..................................................................................... 105

4.5.5 Wind Measurements for VNTRA .............................................................................................. 107

4.5.6 Measurement Statistics for VNTRA .......................................................................................... 107 4.6 Comparison of Monthly Averages ............................................................................................. 112

5 Estimation of long-term best estimate (P50) for GHI & DNI and corresponding uncertainties .................................................................................................................... 114

5.1 Methodology ............................................................................................................................. 114

5.2 VNHAN ..................................................................................................................................... 115 5.2.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement

data ........................................................................................................................................... 115

5.2.2 Determination of long-term average ......................................................................................... 116 5.2.2.1 GHI ........................................................................................................................................... 117 1.1.1.1 DNI ............................................................................................................................................ 118

5.2.3 Analysis of uncertainty .............................................................................................................. 119

5.2.4 Annual cycle of GHI and DNI .................................................................................................... 120 5.3 VNDAN ..................................................................................................................................... 122

5.3.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data ........................................................................................................................................... 122

5.3.2 Determination of long-term average ......................................................................................... 123

5.3.2.1 GHI ........................................................................................................................................... 124

1.1.1.2 DNI ............................................................................................................................................ 125 5.3.3 Analysis of uncertainty .............................................................................................................. 126

5.3.4 Annual cycle of GHI and DNI .................................................................................................... 127

5.4 VNCEH ..................................................................................................................................... 129 5.4.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement

data ........................................................................................................................................... 129 5.4.2 Determination of long-term average ......................................................................................... 130

5.4.2.1 GHI ........................................................................................................................................... 131

1.1.1.3 DNI ............................................................................................................................................ 132

www.suntrace.de

Page 4 of 153

5.4.3 Analysis of uncertainty .............................................................................................................. 133

5.4.4 Annual cycle of GHI and DNI .................................................................................................... 134 5.5 VNSOB ..................................................................................................................................... 136

5.5.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data ........................................................................................................................................... 136

5.5.2 Determination of long-term average ......................................................................................... 137

5.5.2.1 GHI ........................................................................................................................................... 138

1.1.1.4 DNI ............................................................................................................................................ 139 5.5.3 Analysis of uncertainty .............................................................................................................. 140

5.5.4 Annual cycle of GHI and DNI .................................................................................................... 141

5.6 VNTRA ...................................................................................................................................... 143 5.6.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement

data ........................................................................................................................................... 143 5.6.2 Determination of long-term average ......................................................................................... 144 5.6.2.1 GHI ........................................................................................................................................... 145

1.1.1.5 DNI ............................................................................................................................................ 146 5.6.3 Analysis of uncertainty .............................................................................................................. 147 5.6.4 Annual cycle of GHI and DNI .................................................................................................... 148

6 Conclusions ..................................................................................................................... 150 7 Summary .......................................................................................................................... 152 8 Bibliography ..................................................................................................................... 153

www.suntrace.de

Page 5 of 153

Glossary

Term / Abbreviation Definition PV Photovoltaic DHI Diffusive Horizontal Irradiance DNI Direct Normal Irradiance GHI Global Horizontal Irradiance RSI Rotating Shadowband Irradiometer GTI Global Tilted Irradiance SG SolarGis

www.suntrace.de

Page 6 of 153

1 Introduction

After 24 months of continuous measurement operation, this report analyses the measurement results that have been recorded across five solar measurement stations in Vietnam. Measurement stations of type Tier 1 and Tier 2 measure solar irradiance based on Thermopile Pyranometer and Rotating Shadowband Irradiometer. Alongside solar measurements, auxiliary parameters have been recorded to observe on-site climatic conditions. The solar measurement report covers the data recorded by the measurement station from 24 September 2017 through 30 November 2019. The solar measurement results are analysed regarding their relevance to PV and CSP power output.

In October 2018, the international expert Joana Zerbin from Suntrace GmbH visited all five sites for maintenance. During the visit, all instrumentation structures were checked for levelling and functionality, measurement value plausibility, in-situ calibration of pyrheliometer by second pyrheliometer, as well as validation of simultaneous measurements. Moreover, a re-training for both the local staff and the station keeper has been carried out. The re-training focused on daily test routines, verification of the sensor’s proper levelling and error handling. Upon this, a detailed report was provided in October 2018. Very recently, in December 2019, Joana Zerbin visited the sites once more for maintenance and station hand-over. Findings will be considered for analysis.

Prior to presenting the measurement results, detailed information on each site is provided. Furthermore, Suntrace conducted Quality Checks (QCs) and Data Corrections to the recordings. To each station, this report supplies definitions of applied QCs in addition. Solar measurement analysis comprises

• QC results

• Seasonal and diurnal characteristics of GHI and DNI

• Soiling measurement analysis

• Temperature, humidity and (if available) precipitation

• Wind measurements

• Measurement statistics

The report will conclude with a comparison of monthly irradiance for all five sites. Investigated sites will be presented in the following.

www.suntrace.de

Page 7 of 153

2 SITE DESCRIPTION

In this chapter, investigated stations are listed and information on each is given. Preliminary, a sation overview is provided before the location of each station and its equipment are presented.

2.1 Overview of Stations

Table 1: Sites of solar measurement stations in Vietnam within the WB ESMAP programme

Site ID Location ID Type Coordinates Altitude Start Date

VNHAN Hanoi region: Bac Ninh on rooftop of new EVN building

1163 Tier1 21.2013°N

106.0629°E 60 m 1.09.2017

VNDAN Da Nang; on rooftop of EVN/CPC bldg. within the city

1166 Tier2 16.0125°N

108.1865°E 24 m 15.09.2017

VNCEH Central Highlands region on ground before EVN bldg. near hydro spillway

1162 Tier1 12.7535°N

107.8761°E 290 m 26.08.2017

VNSOB Song Binh station location on private house’s rooftop

1167 Tier2 11.2641°N

108.3452°E 62 m 19.09.2017

VNTRA Tri An region near HCMC on top of EVN building; near water spillway

1164 Tier1 11.1024°N

107.0378°E 57 m 29.08.2017

www.suntrace.de

Page 8 of 153

www.suntrace.de

Page 9 of 153

2.1.1 VNHAN

Solar Measurement Station Tier 1 at VNHAN:

Recorded meteorological parameters:

• Global horizontal irradiance (GHI) in W/m2 • Direct normal irradiance (DNI) in W/m2 • Diffuse horizontal irradiance (DHI) in W/m2 • Ambient temperature in °C • Relative humidity in % • Wind speed in m/s ca. 63 m above ground • Wind direction at ca. 63 m above ground • Barometric pressure in hPa • Soiling measurement via 3 reference cells measuring global tilted (10º) irradiance

(GTI) in W/m2

Figure 1: Location of VNHAN. Hanoi in the far background. Image adapted from Google Earth.

www.suntrace.de

Page 10 of 153

Additional information:

• As the roof is 60 m high, the wind mast was installed at a height of 3 m above the rooftop.

• For Tier 1 stations, a 250 Wp photovoltaic panel, batteries and a charge controller are installed.

Table 2: List of installed instruments and measurement sensors at VNHAN

No Instrument Manufacturer Model No. Serial No. 1 Pyrheliometer Hukseflux DR-01 8410 2 Pyranometer 1 (DHI) Hukseflux SR30 4737 3 Pyranometer 2 (GHI) Hukseflux SR30 4738 4 Reference Cell 1 IKS ISET Cell 03110 5 Reference Cell 2 IKS ISET Cell 03091 6 Reference Cell 3 IKS ISET Cell 02672 7 Data logger Blueberry

COMPACT Wilmers Messtechnik 0141 1163

8 Thermo-hygro Sensor E+E EE071 171105000265DA 9 Anemometer Wilmers Messtechnik 0293 1291 10 Wind vane Wilmers Messtechnik 0318 20003589 11 Sun tracker including

shading assembly EKO STR-22G S1513605

Figure 2 Practical training with PECC1, exchange of pyranometers

www.suntrace.de

Page 11 of 153

Figure 3 PECC1 levelling pyranometers Figure 4 PECC1 exchanging pyranometers

Figure 5 Group photo of PECC1 staff and Danang Architectire University after practical training

www.suntrace.de

Page 12 of 153

Figure 6 Maintenance protocol VNHAN 1/3

www.suntrace.de

Page 13 of 153

Figure 7 Maintenance protocol VNHAN 2/3

www.suntrace.de

Page 14 of 153

Figure 8 Maintenance protocol VNHAN 3/3

www.suntrace.de

Page 15 of 153

2.1.2 VNDAN

Solar Measurement Station Tier 2

Recorded meteorological parameters: • Global horizontal irradiance (GHI) in W/m2 • Direct normal irradiance (DNI) in W/m2 • Diffuse horizontal irradiance (DHI) in W/m2 • Ambient temperature in °C • Relative humidity in % • Wind speed in m/s 27 m above ground • Barometric pressure in hPa • Soiling measurement via 3 reference cells measuring global tilted (10º) irradiance (GTI)

in W/m2

Additional information: • The Reichert RSP4G (RSI) is installed, which is known to be a precise and robust

instrument and is less sensitive to soiling. Contrary to pyrheliometers, which are used in Tier 1 stations and require daily cleaning to reach its outstanding accuracy, it is sufficient to clean such RSI on a weekly basis.

• The Hukseflux SR20 is used at the Tier 2 station, which is equipped with dome heating system to reduce the influence of dew, but is not ventilated

• For Tier 2 stations, a 60 Wp photovoltaic panel, batteries and a charge controller are installed.

• As the roof of the EVN building in Da Nang is 24 m high, a wind mast measuring wind speed was installed at a height of 3 m.

Table 3: List of installed instruments and measurement sensors at VNDAN No Instrument Manufacturer Model No. Serial No. 1 Rotating Shadowband

Irradiometer Reichert RSP-4G 17-01

2 Pyranometer 1 Hukseflux SR20 4504 3 Reference Cell 1 IKS ISET Cell 03083 4 Reference Cell 2 IKS ISET Cell 03082 5 Reference Cell 3 IKS ISET Cell 02665 6 Data logger Blueberry

COMPACT Wilmers Messtechnik 0141 1166

7 Thermo-hygro Sensor E+E EE071 17110500023799 8 Anemometer Wilmers Messtechnik 0293 1294

www.suntrace.de

Page 16 of 153

Figure 10 Tier 2 station at VNDAN

Figure 9: Location of VNDAN in the centre of the coastal city of Da Nang, Central Vietnam. Image adapted from Google Earth.

www.suntrace.de

Page 17 of 153

Figure 11 Tier 2 station at VNDAN Figure 12 Exchange opf pyranomter SR20 with re-calibrated SR20

Figure 13 Levelling of RSP Figure 14 New re-calibrated SR20

Figure 15 Wind mast with anemometer in Danang Figure 16 Levelled SR20

www.suntrace.de

Page 18 of 153

Figure 17 Maintenance protocol VNDAN 1/3

www.suntrace.de

Page 19 of 153

Figure 18 Maintenance protocol VNDAN 2/3

www.suntrace.de

Page 20 of 153

Figure 19 Maintenance protocol VNDAN 3/3

www.suntrace.de

Page 21 of 153

2.1.3 VNCEH

Solar Measurement Station Tier 1

Recorded meteorological parameters: • Global horizontal irradiance (GHI) in W/m2 • Direct normal irradiance (DNI) in W/m2 • Diffuse horizontal irradiance (DHI) in W/m2 • Ambient temperature in °C • Relative humidity in % • Wind speed in m/s 10 m above ground • Wind direction at 10 m above ground • Barometric pressure in hPa • Soiling measurement via 3 reference cells measuring global tilted (10º) irradiance (GTI) in

W/m2 Additional Information: • For Tier 1 stations, a 250 Wp photovoltaic panel, batteries and a charge controller are

installed.

Table 4: List of installed instruments and measurement sensors at VNCEH

No Instrument Manufacturer Model No.

Serial No.

1 Pyrheliometer Hukseflux DR02 9219 2 Pyranometer 1 (DHI) Hukseflux SR30 4733 3 Pyranometer 2 (GHI) Hukseflux SR30 4734 4 Reference Cell 1 IKS ISET

Cell 03081

5 Reference Cell 2 IKS ISET Cell

03085

6 Reference Cell 3 IKS ISET Cell

03087

7 Data logger Blueberry COMPACT

Wilmers Messtechnik 0141 1162

8 Thermo-hygro sensor E+E EE071 1711050002481361 9 Anemometer Wilmers Messtechnik 0293 1289 10 Wind vane Wilmers Messtechnik 0218 20003588 11 Sun tracker including shading

assembly EKO STR-

22G S15136.06

www.suntrace.de

Page 22 of 153

Figure 21 Sun tracker at VNCEH Figure 22 Thermo-hygro sensor at VNCEH

Figure 20: Location of VNCEH. Image adapted from Google Earth.

www.suntrace.de

Page 23 of 153

Figure 23 New and old pyranometer 2 (GHI)

Figure 24 Levelling of the re-calibrated pyrheliometer

Figure 25 New and old pyranometer 1 (DHI)

Figure 26 Levelled re-calibrated pyranometers

www.suntrace.de

Page 24 of 153

Figure 27 Maintenance protocol VNCEH (1/3)

www.suntrace.de

Page 25 of 153

Figure 28 Maintenance protocol VNCEH (2/3)

www.suntrace.de

Page 26 of 153

Figure 29 Maintenance protocol VNCEH (3/3)

www.suntrace.de

Page 27 of 153

2.1.4 VNSOB

Solar Measurement Station Tier 2

Recorded Meteorological Parameters: • Global horizontal irradiance (GHI) in W/m2 • Direct normal irradiance (DNI) in W/m2 • Diffuse horizontal irradiance (DHI) in W/m2 • Ambient temperature in °C • Relative humidity in % • Wind speed in m/s 10 m above ground • Wind direction 10 m above ground • Barometric pressure in hPa • Soiling measurement via 3 reference cells measuring global tilted (10º) irradiance (GTI)

in W/m2 • Rain gauge (pluviometer) measuring rain precipitation in mm.

Additional Information: • The Reichert RSP4G (RSI) is installed, which is known to be a precise and robust

instrument and is less sensitive to soiling. Contrary to pyrheliometers, which are used in Tier 1 stations and require daily cleaning to reach its outstanding accuracy, it is sufficient to clean such RSI on a weekly basis.

• The Hukseflux SR20 is used at the Tier 2 station, which is equipped with dome heating system to reduce the influence of dew, but is not ventilated

• For Tier 2 stations, a 60 Wp photovoltaic panel, batteries and a charge controller are installed.

• As the roof of the private house near Song Binh is 4 m high, the wind mast was installed at a height of 6 m above rooftop.

Table 5: List of installed instruments and measurement sensors

No Instrument Manufacturer Model No. Serial No. Comments 1 Rotating

Shadowband Irradiometer

Reichert RSP-4G 17-02 Previous installed 17-03 was replaced

2 Pyranometer 1 Hukseflux SR20-D2 4505 Replaced with a re-calibrated SR20

3 Reference Cell 1

IKS ISET Cell 02623 -

4 Reference Cell 2

IKS ISET Cell 03089 -

www.suntrace.de

Page 28 of 153

No Instrument Manufacturer Model No. Serial No. Comments 5 Reference Cell

3 IKS ISET Cell 02759 -

6 Data logger Blueberry COMPACT

Wilmers Messtechnik

0141 1167 -

7 Thermo-hygro Sensor

E+E EE071 17110500025733 -

8 Anemometer Wilmers Messtechnik

0293 1290 -

9 Wind vane Wilmers Messtechnik

0318 20002437 -

10 Corrosion measurement kit

Fraunhofer ISE

Samples: S155-157, Z155-157, A155-157, C155-157

Was dismantled for assessment.

11 Rain Gauge Young 52203 TB13883 -

Figure 30: Location of VNSOB, east of Sông Binh. Image adapted from Google Earth.

www.suntrace.de

Page 29 of 153

Figure 31 Wind mast at VNSOB Figure 32 Tier 2 station at VNSOB

Figure 33 Rain gauge and Tier 2 station in the back ar VNSOB

www.suntrace.de

Page 30 of 153

Figure 34 Rain gauge from above at VNSOB Figure 35 Check if tipping bucket is clean

Figure 36 Exchange of re-calibrated SR20

Figure 37 Levelled SR20

Figure 38 Levelled RSP

www.suntrace.de

Page 31 of 153

Figure 39 Maintenance protocol at VNSOB (1/3)

www.suntrace.de

Page 32 of 153

Figure 40 Maintenance protocol at VNSOB (2/3)

www.suntrace.de

Page 33 of 153

Figure 41 Maintenance protocol at VNSOB (3/3)

www.suntrace.de

Page 34 of 153

2.1.5 VNTRA

Solar Measurement Station Tier 1

Recorded Meteorological Parameters: • Global horizontal irradiance (GHI) in W/m2 • Direct normal irradiance (DNI) in W/m2 • Diffuse horizontal irradiance (DHI) in W/m2 • Ambient temperature in °C • Relative humidity in % • Wind speed in m/s ca. 13 m above ground • Wind direction at ca. 13 m above ground • Barometric pressure in hPa • Soiling measurement via 3 reference cells measuring global tilted (10º) irradiance (GTI)

in W/m2 Additional observations: • As the EVN building’s roof is ca.10 m high, a 3 m high wind mast was installed above

the rooftop. • For Tier 1 stations, a 250 Wp photovoltaic panel, batteries and a charge controller are

installed.

Table 6: List of installed instruments and measurement sensors

No Instrument Manufacturer Model No.

Serial No. Comments

1 Pyrheliometer Hukseflux DR-02 9220 Replaced with re-calibrated pyrheliometer.

2 Pyranometer 1 (DHI) Hukseflux SR30 4735 Replaced with re-calibrated pyranometer.

3 Pyranometer 2 (GHI) Hukseflux SR30-01 4736 Replaced with re-calibrated pyranometer.

4 Reference Cell 1 IKS ISET Cell

02687 -

5 Reference Cell 2 IKS ISET Cell

02691 -

www.suntrace.de

Page 35 of 153

No Instrument Manufacturer Model No.

Serial No. Comments

6 Reference Cell 3 IKS ISET Cell

02683 -

7 Data logger Blueberry COMPACT

Wilmers Messtechnik

0141 1164 -

8 Thermo-hygro Sensor

E+E EE071 171105000249E8 -

9 Anemometer Wilmers Messtechnik

0293 1292 -

10 Wind vane Wilmers Messtechnik

0318 20003590 -

11 Sun tracker including shading assembly

EKO STR-22G

S1510788 -

12 Corrosion measurement

Fraunhofer ISE Samples: S152-154, Z152-154, A152-154, C152-154

Was dismantled for assessment.

Figure 42: Location of VNTRA. The station is located just outside of the town of Vinh An. Image adapted from Google Earth.

www.suntrace.de

Page 36 of 153

Figure 43 Wind mast at VNTRA Figure 44 Sun tracker at VNTRA

Figure 45 Soiling Measurement System at VNTRA

www.suntrace.de

Page 37 of 153

Figure 46 Tier 1 station at VNTRA Figure 47 Re-calibrated pyranometer

Figure 48 Levelled re-calibrated pyrheliometer Figure 49 Re-calibrated pyrheliometer

www.suntrace.de

Page 38 of 153

Figure 50 Levelled re-calibrated pyranometer Figure 51 Levelled re-calibrated pyranometer

Figure 52 Tier 1 station at VNTRA

www.suntrace.de

Page 39 of 153

Figure 53 Maintenance protocol VNTRA (1/3)

www.suntrace.de

Page 40 of 153

Figure 54 Maintenance protocol VNTRA (2/3)

www.suntrace.de

Page 41 of 153

Figure 55 Maintenance protocol VNTRA (3/3)

www.suntrace.de

Page 42 of 153

2.1.6 Handover of the stations

With the end of the 2-year solar measurement campaign, the measurement station of VNDAN has been handed over to Danang Architecture University for research purposes. The other 4 stations have been handed over to Power Engineering Consulting JSC 1 (PECC1) for their solar power projects as well as for scientific research for PECC1 engineers. PECC1 engineers and Danang Architecture University teachers and students received a comprehensive training on the installation, operation and maintenance of the measurement stations from Suntrace in Hanoi. Also, a practical training was given at the station VNHAN.

Figure 56 Handover ceremony at PECC1 in Hanoi

www.suntrace.de

Page 43 of 153

Figure 57 Handover ceremony at PECC1 in Hanoi

www.suntrace.de

Page 44 of 153

3 Quality Checks

The solar measurement report covers the data recorded by the solar measurement station from 24 September 2017 to 30 November 2019. Assessing the quality of measurements is a crucial aspect of reputable data analysis. Both the solar radiation data and the auxiliary measurements have been checked thoroughly. The quality check definitions used for this data quality assessment will be first explained in this chapter. A list of the quality checks performed, including additional information on the tests, is shown in Table 7.

Table 7: List of quality checks performed; *flags are being added up, if more than one test is positive at a certain time step

Name Description Data Types (solar and/or

auxiliary)

Value in Data File*

Missing Value Identification of missing measurements Both 1e20 Timeshift Error Check for wrong time setting during data

recording Both 1e18

Below Physical Limit

Check if value is below physical limit Both 1e16

Above Physical Limit

Check if value is above physical limit Both 1e15

Tracking/Shading Error

Checks for significant deviations between solar components likely related to tracking/shading errors

Solar 1e12

High Gradient Check for unphysically strong deviations in a certain time interval

Auxiliary 1e12

Cleaning Flag active during site cleanings Both 1e10 2 Component Test

Checks whether Diffusive Horizontal Irradiance (DHI) exceeds G(lobal)HI

Solar 1e7

3 Component Test

Compares GHI measurements with sum of DHI and DNI (on horizontal plane)

Solar 1e6

Corrected Flags noting manual changes Both 1e5 Clear Sky Limit Checks on whether clear-sky radiation is

exceeded Solar 1e4

Correct Flag to denote that the data is assumed to be absolutely correct

Both 1e0

For high quality data assessments of solar radiation measurements, another important aspect is to check whether the recommended cleaning of solar radiation sensors was performed regularly. The flagging performed based on the cleaning events (recorded via a service button) is defined in Table 8.

www.suntrace.de

Page 45 of 153

Table 8: List of cleaning flags

Name Description Value in Data File* Insufficient Cleaning Last cleaning event at least 6 days

ago 1e2

Substandard cleaning

Last cleaning event at least 4 days ago, but less than 6

1e1

Recent cleaning Last cleaning event less than 4 days go

1e0

The results of the quality checks, radiation and auxiliary measurements will be illustrated and discussed for the station in the following chapters.

www.suntrace.de

Page 46 of 153

4 Solar Measurement Results

In this chapter, solar measurement results are presented for each of the five stations; VNHAN, VNDAN, VNCEH, VNSOB and VNTRA. Results provided cover

• QC results

• Seasonal and diurnal characteristics of GHI and DNI

• Soiling measurement analysis

• Temperature, humidity and (if available) precipitation

• Wind measurements

• Measurement statistics

Since only station VNSOB has been equipped with a rain gauge, precipitation measurements can only be provided for this station.

4.1 VNHAN

4.1.1 Quality Checks for VNHAN

The results of the quality checks for VNHAN are illustrated in Error! Reference source not found.. Each horizontal bar contains the relative number of flags identified for a certain parameter. Due to different flag definitions among certain parameter groups, the figure is separated into three parts: radiation parameters, auxiliary parameters and cleaning. In the bar plot of radiation parameters, it can be seen that around 40% to 50% of measurements are flagged as corrected. The reason is that radiation measurements performed by thermopile-based instruments often show negative values during the night. This behaviour is a consequence of the instrument as it emits more long-wave radiation than it receives during night. This effect is caused by inverted temperatures between ambient air and instrument: At night, the measurement device is generally warmer than the effective environmental temperature. This leads to values mostly in the range of –1 to –4 Watts per square meter. This effect cannot be observed in solar power plants, e. g. photovoltaic. Therefore, negative values are set to 0. Consequently, the observed high percentage of corrected data is not a sign of poor data quality. Instead, the data is, overall, of high quality as there are no major gaps in the dataset (not counting periods flagged due to cleaning) or unusual values of any solar radiation sensor during daytime. Affecting thermopile pyranometer measurements, the effect of negative values during nighttime, furthermore, leads to DNI values below the physical mininum. For the greater part, reference cells to estimate the impact by soiling have been cleaned sufficiently i.e. according to the agreed cleaning schedule. However, during January 2018 to April 2018 and July 2018 to October 2018 the cells were cleaned insufficiently i.e. not at all.

www.suntrace.de

Page 47 of 153

Figure 58: Results of Quality Checks for VNHAN. Definitions of flags are given in detail in chapter 3.

4.1.2 Seasonal and Diurnal Characteristics for VNHAN

In this section, seasonal and diurnal variations and characteristics will be discussed for the whole measurement period from September 2017 to November 2019. Figure 59 shows the seasonal variations in radiation for VNHAN, located northeast of Hanoi. Northern Vietnam and the Hanoi region have distinct summer and winter seasons. Nevertheless, high relative variations for all three, GHI, DNI and DHI, can be found during the biennial measurement period. The intermedi-ate but mostly dry winter lasts from November to April. Summer lasts from Mai to October, when hot and humid air predominates the climate. Even though conditions are expected to be clearest during winter, high insolation peaks during summertime, leading to highest irradiance values. On average, GHI values lie well above 150 W/m2 and reach outstandingly high monthly aver-ages of approx. 230 W/m2 in summer. During May and June 2018, as well as during June to September 2019, monthly averages exceed 200 W/m2 Comparing 2018 with 2019, the annual peak reveals to be shifted, indicating high inter-annual variabilities. As Vietnam is affected by

www.suntrace.de

Page 48 of 153

monsoon events, such variations are to be expected. Especially after monsoon events, the cli-mate experiences advection of dry air masses, leading to clear conditions during summer. In Northern Vietnam, monsoon season lasts throughout the country’s meteorological winter, thus, from November to April. During this season, a large fraction of the sky is obscured by clouds on both temporal and spatial behalf. The high percentage in cloud cover causes the observed in-dent in monthly mean insolation. Yet, since more light is scattered by clouds, the total recorded irradiance (GHI) approximately corresponds to the diffusive irradiance (DHI) during winter months. Furthermore, this translates to low direct irradiance (DNI), causing monthly averages to fall down to approx. 10 W/m2 during monsoon season. As it can be observed, DNI minima temporally cover with DHI minima, as do their maxima. Since partially cloudy conditions lower average DNI values and raise average DHI values, this indicates partially cloudy conditions are not common in the Hanoi area.

In addition to seasonal variability, diurnal characteristics of GHI and DNI are of great importance for an optimised PV operation with regard to diurnal cloud formation blocking sunlight. Depend-ing on the location, frequently higher cloud fraction in one half of the day is often expressed in a mean diurnal cycle of irradiation. Arid regions often show higher insolation in the first half of the day, whereas, in the second half of the day, surface heating due to high insolation enables cloud formation by rising thermal plumes (lifting air particles) condensing in the atmosphere. In turn, a location without any clouds during the whole year would be characterized by a defined symmetry of both radiation components GHI and DNI around average solar noon time. Thus, the curvature of diurnal GHI and DHI comprise valid information on local conditions. Figure 60 shows the average day computed from the biennial measurement period at VNHAN. As can be seen, both diurnal GHI and diurnal DNI of measurement year 1 (1 October 2017 to 30 September 2018) correlate very well to measurement year 2 (1 October 2017 to 30 September 2018). This indi-cates intra-annual diurnal stability. Furthermore, the defined symmetry of morning and evening hours in both GHI and DHI show the daily irradiance to not be affected by conditions that depend on the time of day. The fact that both diurnal GHI and diurnal DNI reveal no asymmetries is expected, since temperatures remain above 20°C throughout the year. Thus, atmospheric con-ditions do not favour e.g. fog during morning hours. At noon, average GHI values exceed 500 W/m2, DNI exceeds 200 W/m2 Daily maxima correlate well with solar noon (05:00 UTC) for GHI and DNI for both years. The inter-daily variability (not shown in Figure 60), quantified as the standard deviation for all values to a given time, has the same magnitude for both years.

www.suntrace.de

Page 49 of 153

Figure 59: Monthly averaged irradiation measurements for VNHAN.

Figure 60: Average day for the respective year at VNHAN, based on measurements from 1 October 2017 to 30 September 2018 and 1 October 2018 to 30 September 2019, respectively. The yellow line indicates the average time of solar noon; the time at which the sun is at its highest position.

www.suntrace.de

Page 50 of 153

4.1.3 Soiling Measurements for VNHAN

PV plants are affected by weathering and the deposition of particles on PV panels. This effect is defined as soiling. The high interest in soiling originates from the fact that soiling on PV panels can decrease the power production of the PV plant by approx. 20%. To quantify soiling on-site, the solar measurement station at VNHAN is equipped with three PV reference cells, measuring GTI quoted as HelioScale Soiling Assembly (see Figure 61). Due to different cleaning strategies, the cells are affected by soiling differently. Thus, the effect of soiling can be estimated by comparing the GTI values of each cell. Therefore, the best-practice cleaning schedule is recommended to follow:

• RefCell1: The first cell must always be cleaned during every station cleaning event and serves as a reference.

• RefCell2: The second cell has to be cleaned every three months. • RefCell3: The third cell must never be cleaned manually. Thus, RefCell3 experiences cleaning

only by precipitations and/or wind thrust.

Figure 61: Exemplary HelioScale Soiling Assembly. The image shows the three reference cells that have to be cleaned in different time intervals at site VNHAN. Figure 62 shows the monthly means of GTI measurements from the HelioScale Soiling Assembly, together with its monthly mean soiling loss and monthly accumulated precipitation. The importance of rain measurements for soiling analysis is due to the fact that a sufficient amount of precipitation naturally cleans the sensors. Unfortunately, no precipitation measurements were conducted at VNHAN. Since Reference Cell 1 has to be cleaned during each station visit, it is not surprising that monthly means of GTI lie, on average, above RefCell2 and RefCell3. Soiling losses at Reference Cell 2 and 3 are, hence, calculated as the percentual loss of GTI in reference to Cell 1 (bottom plot in Figure 62). The anti-correlation between the amount of precipitation and the soiling loss can be clearly identified. During the monsoon season (summer; April to October), frequent rain events prevent significant soiling losses. During dry season, however, monthly averaged soiling losses

www.suntrace.de

Page 51 of 153

can reach 2.5%. On monthly scale, soiling losses during the dry season are generally higher for the second measurement year. Most likely, this is related to extended precipitation-free periods during year 2. As a consequence of the performed cleaning schedule, highest soiling losses are expected at Reference Cell 3, which has never been cleaned. However, this tendency cannot be observed throughout the entire duration of the campaign. Reasons therefore are to be found in Vietnam’s climate: An increase in precipitation reduces the detectability of soiling influences, as rainfall will clean all three cells likewise. Vietnam’s weather is heavily influenced by severe rainfall, especially during monsoon season. Thus, monthly averages as presented in Figure 62 do not reveal soiling losses well. Under Vietnam’s climatic conditions, time periods when soiling becomes apparent are commonly only to be observed on sub-monthly (i.e. weekly) scale. Therefore, extended investigations on behalf of soiling will discuss daily GTI values.

Figure 62: Monthly means of GTI measurements at Reference Cells and of soiling losses at VNHAN.

Figure 63 presents the soiling losses for both cells RefCell2 and RefCell3 for the entire measurement period on a daily scale. Revealing more detail, soiling events become apparent during winter (i.e. during dry season). Soiling events characteristically show a strong increase in daily mean soiling loss, followed by a sudden fall-back to approx. 0%. The fall-back is caused by cleaning and consequently a reduction in soiling loss. Cleaning conducted by the station keeper will provoke a fall-back only for RefCell2, whereas rainfall will cause soiling losses at both cells to be reduced. In Figure 63, typical soiling events can be observed during October 2017 and January 2018, as well as during November 2018 and June 2019. Thus, soiling becomes apparent during dry season as expected. However, during the second period (November 2018 and June 2019), soiling losses at RefCell2 show to lie above those at RefCell3. Showing unphysical behaviour, contamination of RefCell3 must be assumed during this period, causing the observed offset. Station maintenance documents were investigated for cause. Unfortunately, no explanation could

www.suntrace.de

Page 52 of 153

be found. Another explanation for this observation could be related to saturation effects in soiling. This phenomenon can be described by the decrease of soiling rates, since the cells were already dirty, or the wind blew away dust from the sensors. However, since no definite cause can be stated, soiling loss estimates are based on the first period, October 2017 and January 2018. During this period, daily soiling losses reach 5% after approx. 15 days if the PV cell is not cleaned (RefCell3). Even rare cleaning (i.e. every three months) can reduce the soiling loss. At RefCell2, for instance, maximum daily soiling losses do not reach 4%. The differences, that still appeared in the measured values, also highlight the complexity of soiling effects on PV systems. During rain season though, daily soiling losses hardly exceeded more than 1 %. Regarding an optimised cost-loss ratio, the detected variability of soiling effects is highly important for further cleaning schedules during PV operation.

Figure 63: Time series of daily soiling impact on reference cells at VNHAN during the two years. Cleaning events by the station keeper are marked as black crosses. As previously mentioned, Vietnam’s climate conditions hold significant precipitation, especially during the rainy season. Such rainfall prevents intense soiling loss. Hence, dry-season measurement periods (between November and April) have been chosen from the data to calculate a daily soiling rate. Representative periods rarely exceed one month. The daily soiling rate serves as an estimate to extrapolate the potential loss of energy production due to insufficient cleaning of the cells. However, soiling is a very complex and time-dependent phenomenon and shows high variability. In fact, daily soiling rates that are representative for longer periods often underestimate daily soiling rates for shorter time scales (less than 15 days) due to saturation effects. For both reference cells, four periods have been considered for VNHAN. The calculated daily soiling rates are listed in Table 9 for each of these periods. In order to provide a better understanding of the applied methodology to derive daily soiling rates, Figure 64 and Figure 65 show daily soiling losses at RefCell2 and RefCell3 and for the four considered periods. By fitting each interval, linear regression represents the value of the daily soiling rate (see corresponding lines in Figure 64 and Figure 65).

www.suntrace.de

Page 53 of 153

Table 9: Daily soiling rates for investigated periods obtained by linear regression

Reference Cell 2 Reference Cell 3

Period from to

Days Daily Soiling Rate Period from to

Days Daily Soiling Rate

2017-10-15 2017-11-01 17 -0.22% 2017-10-15

2017-11-06 22 -0.26%

2017-11-19 2017-12-02 13 -0.19% 2017-11-16

2017-12-03 14 -0.21%

2017-12-28 2018-01-22 25 -0.09% 2017-12-28

2018-01-22 25 -0.11%

2019-01-02 2019-02-09 38 -0.08% 2019-01-02

2019-02-09 38 -0.07%

Averaging the periods given in Table 9 yields the following mean daily soiling rates:

• RefCell 2, mean Daily Soiling Rate: -0.14% • RefCell 3, mean Daily Soiling Rate: -0.16%

This translates to an average daily net loss of 0.15% in solar irradiance uptake. Thus, soiling shows to have timewise significant impact on PV performance.

Figure 64: Soiling Loss in % at RefCell2 for 4 periods during dry season. The corresponding lines are obtained by linear regression and serve as an estimate for the daily soiling rate.

www.suntrace.de

Page 54 of 153

Figure 65: Soiling Loss in % at RefCell3 for 4 periods during dry season. The corresponding lines are obtained by linear regression and serve as an estimate for the daily soiling rate.

4.1.4 Temperature and Humidity for VNHAN

Besides the direct dependency of PV power output on solar irradiance, air temperature plays a fundamental role in PV operation: Affecting the PV system’s performance, PV power output directly depends on temperature. Thus, air temperature is of further interest in the assessment of PV power plant planning and is recorded alongside relative humidity and solar irradiance measurements. Monthly averages are presented in Figure 66.

www.suntrace.de

Page 55 of 153

Figure 66: Monthly temperature (upper plot) and monthly relative humidity (bottom plot). The shaded areas show minimum and maximum daily averages to the corresponding variables for the given months to illustrate intra-monthly variabilities. Shaded areas neighbouring monthly averaged temperature and relative humidty show the minimum and maximum values of daily averages within one month. Its spread illustrates the intra-monthly variability.

www.suntrace.de

Page 56 of 153

Northern Vietnam is characterised by temperate climatic conditions; comprising hot summers and dry winters. VNHAN is affected by hot and very humid climate during the monsoon season aligned with very high precipitation. The higher the temperature of the airmass, the higher its capability to hold water vapour before saturation. Consequently, favouring very humid conditions and, furthermore, holding a large amount of precipitable water in the atmosphere. During the rainy season, the monthly average temperature approximates 30°C. During this period, days with mean temperatures below 25°C have hardly been measured. Highest monthly averages in relative humidity can be found during May to October. During September 2019, the monthly relative humidity peaked above 90%. During 2018, a significant increment in monthly relative humidity during summer months held off. During winter 2018, minimum values in relative humidity drop below 40%, indicating periods with very dry air. Furthermore, this period experienced highest intra-monthly variability in humidity. During summer months, days with an average relative humidity of 100% can also be found in the measurement data. Overall, VNHAN is a very humid site, as monthly means of relative humidity dropped below 80% only in the second dry season of the measurement period. On intra-annual scale, 2019 shows to have been much more humid than 2018. This shows high variability in-between years. Lowest daily mean values of relative humidity below 40 % have been measured during December 2017 and February 2018. Yet, due to monsoon circulation, new airmasses are already advected at the end of the winter, manifesting in rising temperature and days with increasing relative humidity. During dry season, monthly temperature averages reach a minimum of about 17°C. The dry season in the second half of the observation period shows to have been significantly less arid compared to the first period. The annual temperature cycle, however, shows to have been much more stable throughout the observed period.

4.1.5 Wind Measurements for VNHAN

Next to solar measurement instruments, the station at VNHAN is equipped with a wind vane and a cup anemomenter enabling auxiliary measurements. Results of wind measurements are presented as windrose in Figure 67. The windrose reveals winds at VNHAN to be weak and rarely exceed. The average wind direction is south-southeast. Only 5% of the recorded wind gusts (not shown here) exceed 8.0 m/s.

www.suntrace.de

Page 57 of 153

Figure 67: Windrose for VNHAN, showing wind direction, ratio and magnitude of recorded wind data.

4.1.6 Measurement Statistics for VNHAN

In this section, measurement statistics are summarised and presented in detail. During the complete two-year measurement period, mean values of radiation parameters are 150.3 W/m2 for GHI, 73.4 W/m2 for DNI and 97.2 W/m2 for DHI, based on thermopile pyranometer measurements. Table 10 lists mean, minimum and maximum values of measurements at VNHAN for both years related to 1-minute measurements. Even though maxima show to have been significantly higher during year 2, irradiation values are on average constant throughtout the investigated years. Temperature shows to have increased in the second year by 0.9°C. As previously mentioned, the winter of 2017/2018 was unusually dry. Thus, the annual relative humidity during year 2 increased by almost 8%, indicating a high potential for intra-annual variability. Throughout the investigated timespan, the predominant wind direction remained at approx. south-southeast. Since extreme weather can cause significant damage to modules and other equipment, intense rainfall and wind gusts play a crucial role for PV operation. Strong wind gusts, classified as severe gale and correspond to 9 on the Beaufort Scale, have been measured in both years and reached a maximum of 25.7m/s in year 1.

www.suntrace.de

Page 58 of 153

Table 10: Statistics on the whole dataset providing mean, minimum and maximum values of 1-minute values. TH.PYR = Thermopile Pyranometer

PARAMETER UNIT YEAR 1 YEAR 2

MEAN MIN MAX MEAN MIN MAX GHI (TH.PYR) W/m2 149 0 1447 150 0 1376 DNI (TH.PYR) W/m2 72 0 896 71 0 882 DHI (TH.PYR) W/m2 97 0 853 98 0 895 GTI (CLEANED OFTEN) W/m2 151 0 1458 151 0 1354 GTI (CLEANED RARELY) W/m2 151 0 1468 150 0 1355 GTI (NEVER CLEANED) W/m2 150 0 1472 150 0 1360 TEMPERATURE °C 31.7 10.4 55.6 32.6 13.2 55.3 HUMIDITY % 79 20 100 85 31 100 AIR PRESSURE hPa 1005 988 1027 1004 986 1024 WIND SPEED ms-1 2.5 0.0 19.9 2.5 0.0 15.4 WIND GUSTS ms-1 – – 25.7 – – 20.3 WIND DIRECTION °N 166.3 – – 154.3 – –

As already depicted previously, the reference cells serve as quantification of soiling effects on annual average. Both reference cells having not been cleaned at every station visit shows radiation losses in the magnitude of up to 4%. Due to rainfall, soiling estimates could not be completed for all months. As minimum and maximum values are less conservative and can be partly affected by measurement errors, Table 11 shows the 5th and 95th percentile for the main parameters. As irradiation during night is 0 W/m2, only daytime values have been considered for the radiation parameters. For auxiliary parameters, both day and night values have been considered.

Table 11: Percentiles of measurement paramters for the whole two-year measurement period *TH.PYR = Thermopile Pyranometer

PARAMETER UNIT 5-PERCENTILE 95-PERCENTILE GHI (TH.PYR) W/m2 9.7 855.6 DNI (TH.PYR) W/m2 0.0 677.6 DHI (TH.PYR) W/m2 9.6 464.2 TEMPERATURE °C 21.5 48.8 RELATIVE HUMIDITY % 51.3 99.4 WIND GUSTS ms-1 1.1 8.0

www.suntrace.de

Page 59 of 153

Even though highest values of GHI were close to 1450 W/m2 during the measurement period, less than 5 % exceeded 855.6 W/m2 during daytime. The fact that only 5% of the data shows relative humidity values below 51.3% underlies very humid conditions at VNHAN. Moreover, temperature did not drop below 21.5°C in 95% of the measurement period. Due to the impact of windthrow on PV parks, it is worth mentioning that winds exceeded 8.0m/s for 5% of the investigated period. For further details, Table 12 and Table 13 provide monthly averages for both years of the measurement data for the most important parameters regarding PV operation.

www.suntrace.de

Page 60 of 153

Table 12: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 1 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2017-10

0.0 156.7 1155.8

0.0 88.1 893.1

0.0 96.4 673.5

18.1 25.5 34.1

35.6 76.8 100.0

– 2.3 –

– – 16.5

2017-11

0.0 121.4 990.5

0.0 51.0 716.5

0.0 89.1 654.2

13.0 22.0 32.2

28.7 72.5 99.1

– 2.6 –

– – 16.0

2017-12

0.0 95.9 918.6

0.0 63.7 865.5

0.0 60.5 518.1

10.6 17.4 25.9

19.5 70.1 100.0

– 2.3 –

– – 16.0

2018-01

0.0 71.4 958.0

0.0 11.42 703.3

0.0 63.6 612.2

8.0 17.5 27.6

30.8 81.6 100.0

– 2.3 –

– – 17.5

2018-02

0.0 84.5 909.9

0.0 15.0 679.9

0.0 73.9 521.2

10.3 17.1 26.2

25.3 73.8 100.0

– 2.2

– – 11.8

2018-03

0.0 127.1 1057.7

0.0 45.1 736.2

0.0 94.5 697.1

14.6 22.2 29.4

28.8 82.8 100.0

– 2.7 –

– – 18.1

2018-04

0.0 124.3 1238.7

0.0 31.9 740.5

0.0 99.5 853.3

15.1 23.8 32.3

32.1 84.0 99.6

– 2.8 –

– – 18.2

2018-05

0.0 220.8 1287.0

0.0 125.8 885.8

0.0 125.4 827.0

22.3 28.8 37.8

47.2 82.6 99.5

– 3.1 –

– – 20.3

2018-06

0.0 214.0 1383.1

0.0 117.0 892.5

0.0 126.4 849.0

22.3 30.2 38.5

40.2 78.0 98.0

– 2.5 –

– – 25.7

2018-07

0.0 192.5 1446.5

0.0 101.3 895.7

0.0 114.8 756.5

24.2 29.6 39.7

42.1 80.2 96.2

– 2.3 –

– – 19.7

2018-08

0.0 188.6 1343.0

0.0 92.0 846.3

0.0 115.2 780.9

23.7 28.9 37.2

54.5 84.3 95.6

– 2.2 –

– – 22.9

2018-09

0.0 185.9 1219.1

0.0 122.9 880.6

0.0 99.0 795.8

22.5 28.5 36.8

49.6 79.5 95.4

– 2.2 –

– – 13.9

www.suntrace.de

Page 61 of 153

Table 13: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 2 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2018-10

0.0 164.6 1038.3

0.0 112.0 846.7

0.0 89.0 717.9

18.7 25.4 33.7

36.9 77.4 94.2

– 2.5 –

– – 14.0

2018-11

0.0 141.4 1113.8

0.0 95.3 882.1

0.0 84.0 668.3

15.0 23.7 32.2

31.0 78.2 93.8

– 2.2 –

– – 16.2

2018-12

0.0 107.5 928.1

0.0 52.4 792.2

0.0 78.3 616.0

7.9 18.8 30.4

49.4 80.3 93.6

– 2.6 –

– – 14.3

2019-01

0.0 74.4 983.9

0.0 15.2 869.1

0.0 65.3 589.3

9.7 17.1 27.2

54.8 82.5 94.4

– 2.1 –

– – 14.8

2019-02

0.0 108.1 1018.0

0.0 41.0 868.1

0.0 81.0 612.0

13.2 21.7 31.1

54.9 86.2 94.5

– 2.8 –

– – 15.1

2019-03

0.0 91.2 892.6

0.0 16.3 606.0

0.0 78.6 620.7

14.9 22.0 30.6

55.3 86.0 96.1

– 2.4 –

– – 13.6

2019-04

0.0 139.9 1051.8

0.0 41.9 758.1

0.0 104.2 774.8

18.4 26.7 36.6

59.4 89.9 98.7

– 2.9 –

– – 18.9

2019-05

0.0 147.9 1320.6

0.0 49.0 731.8

0.0 107.2 895.3

21.6 27.6 39.9

49.8 89.1 100.0

– 2.6 –

– – 16.3

2019-06

0.0 207.0 1376.0

0.0 86.9 795.0

0.0 135.8 777.5

22.8 30.8 39.4

55.3 88.2 100.0

– 2.9 –

– – 18.3

2019-07

0.0 190.4 1153.5

0.0 93.1 766.7

0.0 113.6 690.0

24.4 30.8 39.4

54.9 88.0 100.0

– 2.7 –

– – 17.2

2019-08

0.0 200.2 1321.7

0.0 105.6 833.1

0.0 114.7 770.4

24.7 29.6 39.3

55.4 92.8 100.0

– 2.2 –

– – 20.3

2019-09

0.0 221.7 1321.7

0.0 157.7 855.6

0.0 101.4 711.4

22.3 29.2 38.3

40.7 83.7 100.0

– 2.5 –

– – 17.2

www.suntrace.de

Page 62 of 153

4.2 VNDAN

4.2.1 Quality Checks for VNDAN

The results of the quality checks for VNDAN are illustrated in Figure 68. Each horizontal bar contains the relative number of flags identified for a certain parameter. Due to different flag definitions among certain parameter groups, the figure is separated into three parts: radiation parameters, auxiliary parameters and cleaning. From the bar plots in solar measurements, it can be seen that approx. 50% of data is flagged as corrected for most parameter. For DNI and DHI only approx. 25% remain uncorrected. At VNDAN i.e. at Tier 2-stations, both a thermopile pyranometer and and a RSI are installed. Solar irradiance at the RSP is recorded with a Silicium-based (SI) pyranometer, which comprises a similar principle as PV cells. For VNHAN i.e. Tier 1 stations, we mentioned a night-time effect leading to negative values, which must be considered unphysical. For the RSP, no such effect occurs. However, it does for the Thermopile Pyranometer. Quality-Check procedures applied to the dataset flag and correct such negative values, explaining a high percentage in corrected data. Furthermore, GHI is measured based on two principles: a SI-based and a Thermopile Pyranometer. Each is sensitive to a different spectrum. Advanced QCs correct this difference in GHI induced by the sensors sensitivity and, hence, further increasing the percentage of corrected data. Consequently, the observed high percentage of corrected data is not a sign of poor data quality. Instead, the data is, overall, of high quality as there are major data gaps (not counting periods flagged due to cleaning) or unusual values of any solar radiation sensor during daytime. For the greater part, reference cells to estimate the impact by soiling have been cleaned suffieciently i.e. according to the agreed cleaning schedule. However, in January 2019 and November 2019, the cells were cleaned insufficiently i.e. not at all. Yet, during year 1, the station was cleaned on only 102 days. During year 2, the cleaning schedule gradually improved, providing overall sufficient cleaning during the second measurement period. Insufficient cleaning will reduce data quality of all parameters. Main periods of insufficient cleaning are from February 2018 to April 2018 and July 2018 to October 2018. This is why all Pyranometer meas-urements from 1 March of 2018 until 25 October 2018 had to be corrected with the aid of GHI and DHI measurements recorded by the Thermopile Pyranometer, since they are less suscepti-ble to soiling. For auxiliary parameters, we observe a high data quality as mostly all values are considered as correct. Only some outliers for the temperature measurements of the data logger can be ob-served, leading temperature gradients classified above physical limit. But as the logger tempera-ture is of no other use than for reference, no relevant issue is present due to that.

www.suntrace.de

Page 63 of 153

Figure 68: Results of Quality Checks for VNDAN. Definitions of flags are given in detail in chapter 3.

4.2.2 Seasonal and Diurnal Characteristics for VNDAN

In this section, seasonal and diurnal variations and characteristics will be discussed for the whole measurement period from September 2017 to November 2019. Figure 69 shows the seasonal variations in radiation for VNHAN in central Vietnam. The monthly averages indicate a strong inner annual variability. On average, the clearest months are from April through Septem-ber. Especially during May 2018 and June 2019 monthly averaged DNI approximate 250 W/m2, indicating very clear conditions during these months. Accordingly, GHI shows highest monthly average values above 270 W/m2. In May 2018, GHI even exceeds 250 W/m2. Furthermore, May 2018 shows relatively low DHI of approx. 80 W/m2. During months with little cloud coverage, DNI is much higher than DHI since less clouds scatter sunlight and more direct radiation reaches the ground. For both years, such months are to be found during summer. For GHI, the seasonal variation reveals a twofold distribution during summer: During springtime (March and

www.suntrace.de

Page 64 of 153

April), as well as towards mid/end summer (August and September), monthly GHI values reaches its annual max at approx. 100 W/m2 for both years. The two peaks for one summer are interrupted by a significant indent during mid-summer. This seasonal behaviour can be ob-served for both 2018 and 2019. Regarding the entire biannual period, September 2018 shows to have been exceptionally sunny as indicated by its high DNI average. The effect of more hu-mid conditions due to the monsoon season can be seen in the irradiance measurements from October 2017 through January 2018, as well as November 2018 through January 2019. During the rainy season, GHI and DNI significantly fall back. Compared this decrease in GHI and DNI, the decrease in DHI is hampered. Thus, this period is dominated by clouds, causing DHI be higher than DNI.

Figure 69: Monthly averaged irradiation measurements for VNDAN. In addition to seasonal variability, diurnal characteristics of GHI and DNI are of great importance for an optimised PV operation with regard to diurnal cloud formation blocking sunlight. Depending on the location, frequently higher cloud fraction in one half of the day is often expressed in a mean diurnal cycle of irradiation. Arid regions often show higher insolation in the first half of the day, whereas, in the second half of the day, surface heating due to high insolation enables cloud formation by rising thermal plumes (lifting air particles) condensing in the atmosphere. In turn, a location without any clouds during the whole year would be characterized by a defined symmetry of both radiation components GHI and DNI around average solar noon time. Thus, the curvature of diurnal GHI and DHI comprise valid information

www.suntrace.de

Page 65 of 153

on local conditions. Figure 70 shows the average day computed from the biennial measurement period at VNDAN. As can be seen, both diurnal GHI and diurnal DNI of measurement year 1 (1 October 2017 to 30 September 2018; indicated by dark colours) lie below year 2 (1 October 2017 to 30 September 2018; indicated by light colours). This shows intra-annual diurnal variability. Furthermore, the distinct asymmetry of morning and evening hours in both GHI and DHI show the daily irradiance to be affected by conditions that depend on the time of day. Espe-cially for DNI, it can be seen that the values are not symmetrically distributed around the aver-age solar noon time; represented by the yellow line. Maximum GHI values were reached in early afternoon. In general, DNI is higher from 05:00 to 08:00 UTC DNI than from 02:00 to 05:00 UTC. This can be explained by the humid conditions. As typical for humid regions, foggy condi-tions frequently appear during morning hours. Since diffusive radiation increased due to clouds, the decrease of GHI due to lower direct solar irradiance is partly compensated. Thus, the dis-tinct asymmetry predominantly affects DNI. Stated discussion concerning diurnal symmetry are valid for both years. At early noon, average GHI values reach 650 W/m2 (biannual average). At late afternoon, DNI reaches approx. 350 W/m2. Thus, daily maxima do not correlate well with solar noon (05:00 UTC). The inter-daily variability (not shown in Figure 70), quantified as the standard deviation for all values to a given time, has the same magnitude for both years.

Figure 70: Average day for the respective year at VNDAN, based on measurements from 1 October 2017 to 30 September 2018 and 1 October 2018 to 30 September 2019, respectively. The yellow line indicates the average time of solar noon; the time at which the sun is at its highest position.

www.suntrace.de

Page 66 of 153

4.2.3 Soiling Measurements for VNDAN

An introduction to the principles of soiling is given in section 4.1.3, where soiling estimates for VNHAN are presented. Analogue, soiling investagations are conducted for VNDAN. The upper plot in Figure 71 shows monthly Global Tilted Irradiance (GTI) over time for Reference Cell 1 (hereinafter RefCell1) to Reference Cell 3 (hereinafter RefCell3). Based on the normalised difference to RefCell1 (cleaned at every stations visit), monthly soiling losses are shown in the bottom plot for RefCell2 and RefCell3. Since RefCell1 has to be cleaned during each station visit, it is not surprising that monthly means of GTI lie, on average, above RefCell2 and RefCell3. During the monsoon season (summer; April to October), frequent rain events prevent significant soiling losses. During dry season, monthly averaged soiling losses can reach 3.0%. Maximum soiling losses were recorded in May 2018 and April 2019. On annual scale, soiling losses during the dry season are generally higher for the second measurement year. Most likely, this is related to extended precipitation-free periods during year 2. As expected, soiling losses are higher for RefCell3 than RefCell2. This shows the impact and significance of cleaning. This observation can be seen for most months during the entire observation period. For the months October to December for both 2017 and 2018, soiling losses at RefCell2 and RefCell3 are approximately identical. This is to be explained by rainfall, as precipitation will clean all three cells equally. However, precipitation has not been recorded at VNDAN. To verify such interpretations and advance soiling analysis, the installation of rain gauges is recommended. Regarding an optimised cost-loss ratio, the detected variability of soiling effects is highly important for further cleaning schedules during PV operation.

Figure 71: Monthly means of GTI measurements at Reference Cells and of soiling losses at VNDAN.

www.suntrace.de

Page 67 of 153

Figure 72 presents the soiling losses for both cells RefCell2 and RefCell3 for the entire measurement period on a daily scale. Revealing more detail, soiling events become apparent during winter (i.e. during dry season). Soiling events characteristically show a strong increase in daily mean soiling loss, followed by a sudden fall-back to approx. 0%. The fall-back is caused by cleaning and consequently a reduction in soiling loss. Cleaning conducted by the station keeper will provoke a fall-back only for RefCell2, whereas rainfall will cause soiling losses at both cells to be reduced. In Figure 72, typical soiling events can be observed during January and June 2018 and 2019. Thus, soiling becomes apparent towards the end of winter time. Distinct soiling can only be observed for short periods of time (commonly < 1 month), indicating frequent re-oocuring rainfall. During summer, only minor soiling impact can be observed and requires precipitation measurements for a detailed analysis. On daily scale, maximum soiling losses exceed 5%. Black crosses indicating cleaning events, frequent and sufficient station maintenance can be reported.

Figure 72: Time series of daily soiling impact on reference cells at VNDAN during the two years. Cleaning events by the station keeper are marked as black crosses. As previously mentioned, Vietnam’s climate conditions hold significant precipitation, especially during the rainy season. Such rainfall prevents intense soiling loss. Hence, dry-season measurement periods (between January and June of both years) have been chosen from the data to calculate a daily soiling rate. Representative periods rarely exceed one month. The daily soiling rate serves as an estimate to extrapolate the potential loss of energy production due to insufficient cleaning of the cells. However, soiling is a very complex and time-dependent phenomenon and shows high variability. In fact, daily soiling rates that are representative for longer periods often underestimate daily soiling rates for shorter time scales (less than 15 days) due to saturation effects. At VNDAN, representative periods are short. Thus, saturation effects come do not show and are not expected to dampen soiling estimates. For both reference cells, four periods have been considered for VNDAN. The calculated daily soiling rates are listed in Table 14 for each of these periods. Section 4.1.3 provided figures, that illustrate the methodology for soiling rate estimation and show fitting applied to data recorded at VNHAN, as conducted in analogue for VNDAN.

www.suntrace.de

Page 68 of 153

Table 14: Soiling rates at VNDAN

Reference Cell 2 Reference Cell 3

Period from to

Days Daily Soiling Rate Period from to

Days Daily Soiling Rate

2018-01-24 2018-02-09 16 -0.14% 2018-01-24

2018-02-09 16 -0.19%

2018-05-16 2018-06-01 16 -0.25% 2018-05-14

2018-06-01 18 -0.18%

2019-02-28 2019-03-18 18 -0.18% 2019-03-02

2019-03-23 21 -0.19%

2019-03-30 2019-04-10

11 -0.21% 2019-04-01 2019-04-28

27 -0.15%

Averaging the periods given yields the following mean daily soiling rates:

• RefCell 2, mean Daily Soiling Rate: -0.20% • RefCell 3, mean Daily Soiling Rate: -0.18%

This translates to an average daily net loss of 0.19% in solar irradiance uptake. Thus, soiling shows to have timewise significant impact on PV performance.

4.2.4 Temperature and Humidity for VNDAN

Besides the direct dependency of PV power output on solar, air temperature plays a fundamen-tal role in PV operation. Affecting the PV system’s performance, PV power output directly de-pends on temperature. Thus, air temperature is of further interest in the assessment of PV power plant planning and is recorded alongside relative humidity and solar irradiance measure-ments. Monthly averages are presented in Figure 73. Here, shaded areas neighbouring monthly averaged temperature and relative huminidty show the minimum and maximum values of daily averages within one month. Its spread illustrates the intra-monthly variability. Central Vietnam is characterised by tropical monsoon climate. Due to the monsoon’s cooling effect, temperatures are to expected to be lower compared to the neighbouring VNDAN that is affected by hot and very humid climate during the monsoon season aligned with very high precipitation. During summer, the monthly average temperature approximates 30°C. During this period, days with mean temperatures below 25°C have hardly been measured. Highest monthly averages in relative humidity can be found from November to January. During these months, average humidity of approx. 87% have been recorded. From the plots below, a anti-correlation between relative humidity and temperature can be observed. Thus, high temperatures go alongside relatively low humidity and vice versa.

www.suntrace.de

Page 69 of 153

Overall, monthly averages did not drop below 70% in humidity and 22°C in temperature. While humidity shows to have followed an approx. identical cycle, the second year of measurements was by 1.3°C warmer compared to year 1.

Figure 73: Monthly temperature (upper plot) and monthly relative humidity (bottom plot). The shaded areas show minimum and maximum daily averages to the corresponding variables for the given months to illustrate intra-monthly variabilities.

4.2.5 Measurement Statistics for VNDAN

In this section, measurement statistics are presented in more detail. During the complete two-year measurement period, mean values of radiation parameters are 188 W/m2 for GHI, 128 W/m2 for DNI and 95 W/m2 for DHI, based on Rotating Shadowband Irradiance (RSI) measurements. Table 15 shows mean, minimum and maximum values at VNDAN for both years of the whole-time span related to 1-minute measurements. First-year mean values are calculated for data from 1 October 2017 to 30 September 2018 and second-year mean values from 1 October 2018 to 30 September 2019, respectively. Please note that VNDAN is neither equipped with a wind vane nor with a rain gauge. Thus, not measurements on wind direction and precipitation are listed.

www.suntrace.de

Page 70 of 153

Table 15: Statistics on the whole dataset providing mean, minimum and maximum values of 1-minute values.

PARAMETER UNIT YEAR 1 YEAR 2

MEAN MIN MAX MEAN MIN MAX GHI (TH.PYR) W/m2 181 0 1397 200 0 1419

GHI (RSI) W/m2 167 0 1372 177 0 1288

DNI (RSI) W/m2 105 0 869 127 0 889

DHI (RSI) W/m2 88 0 1109 84 0 717

GTI (CLEANED OFTEN) W/m2 181 0 1463 199 0 1435

GTI (CLEANED RARELY) W/m2 180 0 1464 199 0 1431

GTI (NEVER CLEANED) W/m2 179 0 1459 197 0 1426

TEMPERATURE °C 28.7 13.8 44.7 30.0 17.0 44.7

HUMIDITY % 81 39 100 80 27 97

AIR PRESSURE hPa 1006 978 1022 1006 990 1019

WIND SPEED ms-1 2.3 0.0 13.3 2.3 0.0 12.9

WIND GUSTS ms-1 – – 19.2 – – 18.5 The statistics reveal measurement year 2 to have been characterised by higher irradiation compared to year 1. This increase becomes apparent for all three; GHI, DNI and DHI. This observation correlates well with the annual mean temperature having been by 1.3°C higher for year 2. Strong wind gusts, classified as “severe gale” and correspond to 9 on the Beaufort Scale, have been measured in both years with a maximum wind of up to 19.2m/s in year 1. Furthermore, annual GTI values show to be affected by soiling Soiling. For both years, soiling resulted in a decrease by 1% in average GTI, when comparing RefCell1 with RefCell3. Previously, soiling rates were calculated and had been estimate to correspond to soiling losses by 0.19% per day. Thus, soiling losses can be considerable high at VNDAN. Due to rainfall, the total loss is hampered, explaining a reduced net loss in annual irradiance uptake by RefCell2 and RefCell3. As minimum and maximum values are less conservative and can be partly affected by measurement errors, Error! Reference source not found. shows the 5th and 95th percentile for the main parameters. As irradiation during night is 0, only daytime values have been considered for the radiation parameters. For auxiliary parameters, day and night values have been considered. Hence, these percentiles correspond to data periods of around 875 hours during which measurement values are below (5th percentile) or above (95th percentile) the threshold.

www.suntrace.de

Page 71 of 153

Table 16: Percentiles of measurement paramters for the whole two-year measurement period

PARAMETER UNIT 5-PERCENTILE 95-PERCENTILE

GHI (TH.PYR) W/m2 12 948

GHI (RSI) W/m2 10 856

DNI (RSI) W/m2 0 738

DHI (RSI) W/m2 9 429

TEMPERATURE °C 23.2 40.0

RELATIVE HUMIDITY % 54.3 93.3

WIND GUSTS m/s 0.8 5.7 Even though highest values of GHI were close to 1400 W/m2 during the measurement period, less than 5 % exceeded 948 W/m2 during daytime. The fact that only 5% of the data shows relative humidity values below 54.3% highlights that the site underlies very humid conditions. Moreover, temperature did not drop below 23.2°C in 95 % of the measurement period. For further details, Table 17 and Table 18 provides monthly averages for year 1 and year 2 of the measurement data for the most important parameters regarding PV operation

www.suntrace.de

Page 72 of 153

Table 17: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 1 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [degC] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] MIN MEAN MAX

WGUST [M/S] MIN MEAN MAX

2017-10

0.0 149.6 1395.3

0.0 72.2 905.5

0.0 99.5 762.4

22.0 26.4 33.4

52.3 82.0 99.3

– 2.3 –

– – 16.1

2017-11

0.0 98.0 1183.2

0.0 49.9 914.4

0.0 66.1 1181.3

19.6 24.8 32.3

54.7 88.7 100.0

– 2.6 –

– – 18.6

2017-12

0.0 83.2 1118.1

0.0 30.7 860.3

0.0 64.2 772.0

16.0 21.9 28.7

49.0 85.6 100.0

– 2.3 –

– – 15.3

2018-01

0.0 102.1 1150.9

0.0 32.2 893.2

0.0 82.7 679.0

16.7 22.1 31.4

41.5 86.7 100.0

– 2.0 –

– – 13.6

2018-02

0.0 164.0 1298.8

0.0 101.0 902.0

0.0 95.3 706.7

15.2 21.4 28.4

53.9 81.9 100.0

– 2.5 –

– – 11.5

2018-03

0.0 185.1 1188.1

0.0 100.8 814.5

0.0 107.1 715.8

18.7 24.1 31.2

50.3 83.0 99.2

– 2.3 –

– – 13.4

2018-04

0.0 243.1 1283.5

0.0 169.6 890.3

0.0 108.0 787.6

18.4 25.8 34.2

49.7 82.7 98.7

– 2.4 –

– – 15.3

2018-05

0.0 277.7 1360.6

0.0 245.4 915.8

0.0 85.5 657.7

23.8 28.7 38.9

42.8 78.0 97.4

– 2.4 –

– – 15.4

2018-06

0.0 219.1 1354.5

0.0 134.4 929.2

0.0 116.4 743.3

22.5 29.7 37.6

40.9 73.3 98.3

– 2.4 –

– – 19.2

2018-07

0.0 198.6 1397.3

0.0 114.9 907.3

0.0 112.0 755.8

24.3 29.3 37.7

41.8 76.2 98.9

– 2.2 –

– – 14.4

2018-08

0.0 208.3 1367.0

0.0 120.1 910.4

0.0 144.7 799.5

24.4 29.8 37.4

39.0 72.1 97.3

– 2.3

– – 15.4

2018-09

0.0 250.5 1392.4

0.0 212.8 919.6

0.0 94.3 732.3

23.4 28.6 36.4

46.3 78.0 96.7

– 2.5 –

– – 14.0

www.suntrace.de

Page 73 of 153

Table 18: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 2 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2018-10

0.0 189.7 1158.4

0.0 142.6 904.5

0.0 92.3 757.6

19.3 26.7 33.7

27.0 80.0 96.4

– 2.2 –

– – 14.9

2018-11

0.0 146.0 1204.0

0.0 102.2 913.6

0.0 82.1 741.1

19.8 25.5 30.9

43.4 82.5 95.8

– 2.2 –

– – 13.6

2018-12

0.0 101.5 1025.3

0.0 55.7 876.3

0.0 69.1 676.6

18.1 23.9 30.5

62.0 86.5 96.8

– 2.5 –

– – 14.5

2019-01

0.0 119.4 1156.81

0.0 62.2 952.6

0.0 81.3 679.0

17.1 22.2 28.3

48.7 85.7 94.5

– 1.9 –

– – 10.6

2019-02

0.0 214.2 1167.0

0.0 186.5 980.8

0.0 87.0 631.1

19.7 25.0 31.8

54.1 82.3 95.8

– 2.2

– – 11.0

2019-03

0.0 203.7 1170.0

0.0 126.9 833.2

0.0 106.7 749.0

21.2 26.1 33.5

54.6 82.6 93.1

– 2.3 –

– – 18.4

2019-04

0.0 254.4 1122.4

0.0 184.6 854.9

0.0 105.1 739.3

22.7 28.0 38.9

45.3 80.6 93.0

– 2.2 –

– – 10.3

2019-05

0.0 261.1 1341.9

0.0 204.2 897.9

0.0 99.7 769.9

22.9 29.5 39.4

45.0 77.4 93.7

– 2.4 –

– – 17.4

2019-06

0.0 272.6 1346.4

0.0 243.4 917.5

0.0 90.2 693.3

24.8 31.1 38.6

39.6 72.4 92.3

– 2.5 –

– – 16.7

2019-07

0.0 234.5 1330.6

0.0 175.5 965.6

0.0 99.4 792.2

23.8 30.5 38.4

41.4 70.9 93.2

– 2.5 –

– – 18.5

2019-08

0.0 211.2 1408.0

0.0 115.5 946.2

0.0 122.3 701.7

22.8 29.1 38.5

41.9 73.1 94.0

– 2.6 –

– – 14.7

2019-09

0.0 192.8 1419.3

0.0 115.2 927.7

0.0 106.0 827.7

22.7 27.7 36.5

52.2 80.8 94.1

– 2.1 –

– – 13.0

www.suntrace.de

Page 74 of 153

4.3 VNCEH

4.3.1 Quality Checks for VNCEH

The results of the quality checks for VNCEH are illustrated in Figure 74. Each horizontal bar contains the relative number of flags identified for a certain parameter. Due to different flag-defi-nitions among certain parameter groups, the figure is separated into three parts: Radiation pa-rameters, auxiliary parameters and cleaning. For radiation parameters, the graph reveals approx. 50% of all measurements are marked as corrected. The reason is that radiation measurements performed by thermopile-based instru-ments often show negative values at night. This behaviour is a consequence of the instrument as it emits more long-wave radiation than it receives because it is generally warmer than the ef-fective environmental temperature at night. It leads to values mostly in the range of –1 to –4 W/m2. This effect cannot be observed for solar power plants, e.g. photovoltaic; thus, values are set to 0. In consequence, the high percentage is not a sign of poor data quality but instead re-sults of this night-time phenomenon. If the irradiance of the thermopile instrument during night is stronger than we normally observe, the values are flagged as “below physical limit”, which can be seen for the thermal pyrheliome-ter. Also, these values are set to be 0 to suspend data degradation. Additionally, approx. 10% of GTI measurements at RefCell2 and RefCell3 are missing. Due to a broken cable, GTI measure-ments are only available as of 21 December 2017 for RefCell3. Due to false cabling, GTI meas-urements at RefCell2 measurements are available as of March 2018. Further issues at VNCEH concern the solar tracker: Station surveillance revealed the sun to not being tracked properly during a period. Consequently, DNI and DHI was not measured correctly. The pyrheliometer didn’t point to the sun and the DHI pyranometer were affected by direct insolation, since the shadowball did not entirely shade off the sun. The malfunction was corrected by replacing the sun sensor of the tracker on 28 October 2019. These malfunctions lead to approx. 3% in DNI and DHI measurements being flagged as missing, as can be seen in the QC results below.

www.suntrace.de

Page 75 of 153

Figure 74: Quality Check results for VNCEH. Definitions are given in section 3.

4.3.2 Seasonal and Diurnal Characteristics for VNCEH

In this section, seasonal and diurnal variations and characteristics will be discussed for the whole measurement period from September 2017 to November 2019. Figure 75 shows the seasonal variations in radiation for VNCEH in the central highlands. On average, clearest months were from February to May in 2018. In 2019, the period characterised by clear conditions was prolonged and reached from February through June. Furthermore, September 2017 shows to have been untypically dry. For these periods, average GHI values lie at approx. 250 W/m2. During the first year of observation, annual GHI averaged at 213 W/m2, which is be-low the annual GHI of 224 W/m2 for year 2. For both years 2018 and 2019, low DHI values during February to May indicate a lower occur-rence in clouds than compared to June to August. Higher DHI values from June to August are a consequence of more clouds in the rainy season, caused by the south-west monsoon.

www.suntrace.de

Page 76 of 153

Convection in the afternoon decreases average DNI values and increases DHI values. Statisti-cally, the rainy season starts in April and lasts until November. According to DHI, this can only partly be identified in our radiation measurements, as it seems that there was a temporal shift in the appearance of the south-west monsoon by two months regarding DNI for both years.

Figure 75: Monthly averaged irradiance measurements for VNCEH. Figure 76 shows the average day computed from the biennial measurement period for VNCEH. As for station VNDAN, both diurnal GHI and diurnal DNI of measurement year 1 (1 October 2017 to 30 September 2018; indicated by dark colours) lie below year 2 (1 October 2017 to 30 September 2018; indicated by light colours) at VNCEH. This shows intra-annual diurnal variability. Furthermore, the distinct asymmetry of morning and evening hours in both GHI and DHI show the daily irradiance to be affected by conditions that depend on the time of day. This asymmetry is related to convection, which causes DNI to decrease as of approx. 07:00 UTC. Since ground temperatures reach their daily maximum at noon, the ascent of thermal plumes is intensified. In turn, this will lead to an increase in cloud formation and, consequently, will lead to the observed reduction in evening DHI. In the Central Highlands, orographic effects are likely to amplify convection.

www.suntrace.de

Page 77 of 153

Figure 76: Average day for the respective year at VNDAN, based on measurements from 1 October 2017 to 30 September 2018 and 1 October 2018 to 30 September 2019, respectively. The yellow line indicates the average time of solar noon. A location withouth any clouds is characterised by a clear symmetry to solar noon.

4.3.3 Soiling Measurements for VNCEH

In this section, the impact of soiling on the PV is assessed by GTI-reference measurements. To follow up of the principles of soiling estimates, please refer to section 4.1.3, where an introduction is given. Analogue, soiling investagations are conducted for VNCEH. Due to broken cables and followed-up false cabling, Error! Reference source not found. presents monthly GTI values (upper plot) and monthly soiling losses (bottom plot) only as of January 2018 i.e. March 2018. To be more precise, the malfunction occurred for both RefCell2 & RefCell3 from 27 August 2017 to 10 March 2018. Even though cables were replaced by 21 December 2017, false cabling caused subsequent errouenous data until March 2018. The upper plot in Error! Reference source not found. shows monthly Global Tilted Irradiance (GTI) over time for RefCell1, RefCell2 and RefCell3. RefCell1 has to be cleaned during each station visit, whereas RefCell2 is cleaned once a month and RefCell3 is never cleaned. Based on the normalised difference to RefCell1, monthly soiling losses are shown in the bottom plot for RefCell2 and RefCell3. Thus, GTI values (upper plot) are expected to be least for RefCell3. Accordingly, soiling losses (lower plot) at RefCell2 are expected to be highest for RefCell3. On

www.suntrace.de

Page 78 of 153

monthly average, this assumption can only be confirmed for periods before May 2018 and between August 2018 and March 2019. As mentioned in previous sections, frequent precipitation hampers soiling estimates at monthly scale. Thus, at monthly scale, soiling losses are expected to become apparent during the dry season. At VNCEH, dry season is expected to last from August to December (Gobin et al. 2016). As can be taken from Error! Reference source not found., the periods do not cover entirely with the expected dry season. Due to annual variability, the rainy season may have occurred earlier that year. However, to varify this, on-site rain measurements are a prerequisite. Yet, the conducted analysis reveals monthly soiling losses of up to 1.5%.

Figure 77: Monthly means of GTI measurements at Reference Cells and of monthly soiling loss. To provide further detail on the soiling analysis, Figure 78 shows daily soiling losses. The black crosses show cleaning undertaken by the station keeper and reveal frequent cleaning. Furthermore, the experts at Suntrace investigated on station maintenance documents and could not find any shortcomings in instrument cleaning. Yet, daily soiling rates have shown to be beyond physically realistic limits. Thus, soiling rates as presented in Figure 78 do not consider data from

• RefCell2 & RefCell3: 06 March to 09 March 2018 • RefCell2 & RefCell3: 17 October 2018 to 20 October 2018 • RefCell2 & RefCell3: 17 November 2018 to 19 November 2018

Since unphysically values are to be observed for both RefCell2 and RefCell3 for the same days, it is likely for ambient conditions to have caused these outliers. However, Figure 78 reveals daily soiling rates to exceed 2.5%.

www.suntrace.de

Page 79 of 153

Figure 78: Time series of daily soiling impact on reference cells at VNCEH during the two years. Cleaning events by the station keeper are marked by black crosses. At VNCEH, only few short periods can be found that show distinct soiling. The methodology of finding such periods is presented in section 4.1.3 and will not be discussed at this point. Selected periods for soiling rate estimation are given in alongside the found daily soiling rate. Please note that chosen periods are considerably short. Due to short timespan, saturation effects do not become apparent. Thus, soiling rates are expected to be high; nonetheless, not incorrect.

Table 19: Daily soiling rates at VNCEH

Reference Cell 2 Reference Cell 3

Period from to

Days Daily Soiling Rate Period from to

Days Daily Soiling Rate

– – – 2018-04-01 2018-04-24 23 -0.08%

2019-03-01 2019-03-21 20 -0.07% 2019-03-01

2019-03-28 18 -0.04%

2019-04-15 2019-04-30 15 -0.05% 2019-04-15

2019-04-30 15 -0.05%

Averaging the periods given yields the following mean daily soiling rates:

• RefCell 2, mean Daily Soiling Rate: -0.06% • RefCell 3, mean Daily Soiling Rate: -0.06%

This translates to an average daily net loss of 0.06% in solar irradiance uptake. Thus, soiling shows to have timewise significant impact on PV performance.

www.suntrace.de

Page 80 of 153

4.3.4 Temperature and Humidity for VNCEH

Climatic conditions of the ambient directly and indirectly influencing PV performance, auxiliary measurements complete extensive assessments on solar resources. Therefore, Figure 79 shows the recorded air temperature and relative humidity that have been recorded alongside solar irradiance measurements. Here, shaded areas neighbouring monthly averaged temperature and relative humidity show the minimum and maximum values of daily averages within one month. Its spread illustrates the intra-monthly variability. Vietnam’s Central Highlands are influenced by equatorial climate leading to average monthly temperatures ranging between 25°C and 30°C. Since Central Vietnam is characterised by tropical monsoon climate, the climate is divided into two distinct seasons: a rainy and a dry season. Due to the monsoon’s cooling effect, air temperature is expected to cool down with the rainy season. Overall, air temperature remains fairly constant throughout the year, showing only little seasonal variations. In contrary, the monthly averaged humidity shows much stronger seasonal variations: From February to April, relative humidty approximates its annual minimum near 62%. From August until October, the monthly humidity reaches its maximum above 80%.

www.suntrace.de

Page 81 of 153

Figure 79: Monthly temperature (upper plot) and monthly relative humidity (bottom plot) at VNCEH. The shaded areas show minimum and maximum daily averages to the corresponding variables for the given months to illustrate intra-monthly variabilities.

www.suntrace.de

Page 82 of 153

4.3.5 Wind Measurements for VNCEH

Next to solar measurement instruments, the station at VNCEH is equipped with a wind vane and a cup anemomenter enabling auxiliary measurements. Results of wind measurements are presented as windrose in Figure 80. The windrose reveals winds at VNCEH to be weak and rarely exceed 6 m/s. The average wind direction is southeast. Only 5% of the recorded wind gusts (not shown here) exceed 7.2 m/s.

Figure 80: Windrose for VNCEH.

4.3.6 Measurement Statistics for VNCEH

In this section, measurement statistics are presented in more detail. During the complete two-year measurement period, mean values of radiation parameters are 218 W/m2 for GHI, 158 W/m2 for DNI and 102 W/m2 for DHI, based on Thermopile Pyranometer measurements. Table 20 shows mean, minimum and maximum values at VNCEH for both years of the whole-time span related to 1-minute measurements. First-year mean values are calculated for data from 1 October 2017 to 30 September 2018 and second-year mean values from 1 October 2018 to 30 September 2019, respectively.

www.suntrace.de

Page 83 of 153

Table 20: Statistics on the whole dataset providing mean, minimum and maximum values of 1-minute values.

PARAMETER UNIT YEAR 1 YEAR 2

MEAN MIN MAX MEAN MIN MAX GHI (TH.PYR) W/m2 213 0 1433 224 0 1450

DNI (TH.PYR) W/m2 146 0 984 172 0 990

DHI (TH.PYR) W/m2 105 0 1167 98 0 863

GTI (CLEANED OFTEN) W/m2 215 0 1470 226 0 1435

GTI (CLEANED RARELY) W/m2 217 0 1593 225 0 1429

GTI (NEVER CLEANED) W/m2 221 0 1464 225 0 1429

TEMPERATURE °C 33.6 19.4 54.2 35.5 21.7 57.4

HUMIDITY % 75 22 99 72 26 95

AIR PRESSURE hPa 979 967 989 979 954 988

WIND SPEED ms-1 2.7 0.0 17.0 2.7 0.0 19.2

WIND GUSTS ms-1 – – 21.3 – – 26.0

WIND DIRECTION °N 157.4 – – 154.6 – –

On average, GHI was 5% higher in the second measurement year. Also, the annual DNI increased by as much as 18%. In contrary, DHI decreased by almost 7%. This goes to show strong intra-annual variability. Reference-cell measurements do not reveal a soiling impact on annual average. However, soiling events could be observed and showed an impact at sub-monthly scale. On annual scale, the soiling impact most likely does not become apparent due to rainfall. Annual temperatures increased by 1.9°C in year 2. Predominant wind direction is south-east. During both years, severe gales have been measured and reached up to 26.0 m/s in the second year. As minimum and maximum values are less conservative and can be partly affected by measurement errors, Table 21 shows the 5th and 95th percentile for the main parameters. As irradiation during night is 0 W/m2, only daytime values have been considered for the radiation parameters. For auxiliary parameters, day and night values have been considered.

From October 2017 to November 2019, less than 5% of GHI measurements exceeded 991 W/m2 during daytime. DHI measurements revealed that 95% of diffusice radiation lie below 462 W/m2 . The fact that only 5% of the data shows relative humidity values below 44.9% highlight that the site underlies very humid conditions. Moreover, temperature did not drop below 16.2°C in 95 % of the measurement period.

www.suntrace.de

Page 84 of 153

Table 21: Percentiles of measurement paramters for the whole two-year measurement period

PARAMETER UNIT 5-PERCENTILE 95-PERCENTILE GHI (TH.PYR) W/m2 17.9 990.7 DNI (TH.PYR) W/m2 0.0 854.4 DHI (TH.PYR) W/m2 17.1 461.5 TEMPERATURE °C 28.9 47.1 RELATIVE HUMIDITY % 44.9 88.9 WIND GUSTS m/s 0.7 7.2

For further details, Table 22 and Table 23 provide monthly averages for both years of the measurement data for the most important parameters regarding PV operation.

www.suntrace.de

Page 85 of 153

Table 22: : Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 1 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2017-10

0.0 196.5 1376.2

0.0 117.8 939.5

0.0 108.3 786.7

19.9 26.6 34.6

45.7 80.7 99.1

– 2.1 –

– – 15.1

2017-11

0.0 156.6 1219.3

0.0 93.3 923.1

0.0 97.1 678.0

20.6 26.1 34.4

51.4 81.1 97.7

– 2.5 –

– – 18.9

2017-12

0.0 152.7 1144.7

0.0 102.5 944.8

0.0 88.5 610.5

17.3 23.9 32.3

45.0 76.4 96.8

– 3.5 –

– – 15.4

2018-01

0.0 190.5 1186.7

0.0 147.0 959.8

0.0 98.0 646.0

18.2 24.8 34.7

44.6 74.3 93.9

– 3.7 –

– – 15.5

2018-02

0.0 248.1 1208.1

0.0 248.5 957.5

0.0 79.1 650.9

13.8 24.6 35.4

29.7 66.8 89.8

– 4.1 –

– – 15.6

2018-03

0.0 254.6 1291.2

0.0 198.5 984.1

0.0 104.2 637.8

19.8 27.1 37.1

25.2 66.3 92.8

– 3.6 –

– – 15.5

2018-04

0.0 272.3 1287.4

0.0 218.8 963.5

0.0 102.6 711.7

20.0 28.8 40.0

22.1 64.1 95.7

– 3.2 –

– – 17.1

2018-05

0.0 260.5 1311.0

0.0 195.3 902.4

0.0 102.9 737.7

22.8 28.2 36.4

39.9 74.5 94.6

– 2.0 –

– – 21.2

2018-06

0.0 202.7 1387.7

0.0 105.3 902.1

0.0 120.5 801.0

22.7 27.3 34.1

50.0 78.5 94.9

– 1.9 –

– – 14.9

2018-07

0.0 194.5 1296.3

0.0 83.7 941.2

0.0 128.2 793.4

22.7 27.0 35.1

48.7 79.6 94.2

– 2.0 –

– – 16.3

2018-08

0.0 196.2 1432.7

0.0 85.1 917.0

0.0 127.5 739.3

22.6 26.6 33.3

55.6 80.5 94.3

– 2.1

– – 19.1

2018-09

0.0 235.8 1389.3

0.0 166.2 941.5

0.0 104.5 1167.0

21.9 26.9 34.3

52.8 80.3 95.2

– 1.8 –

– – 15.2

www.suntrace.de

Page 86 of 153

Table 23: : Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 2 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2018-10

0.0 233.1 1296.2

0.0 189.7 942.8

0.0 94.3 640.4

18.9 27.1 35.7

37.9 72.9 93.6

– 2.8 –

– – 26.0

2018-11

0.0 193.8 1259.1

0.0 174.4 954.2

0.0 81.1 667.5

20.0 26.3 35.1

40.1 73.6 92.6

– 3.1 –

– – 14.4

2018-12

0.0 165.6 1211.5

0.0 128.7 588.4

0.0 82.3 641.9

19.4 25.7 34.0

41.6 74.9 92.4

– 3.1 –

– – 15.7

2019-01

0.0 204.4 1184.4

0.0 195.6 957.5

0.0 80.7 597.4

16.4 24.5 32.3

36.7 70.5 88.9

– 4.1 –

– – 17.1

2019-02

0.0 266.5 1156.5

0.0 290.7 990.1

0.0 66.3 614.8

18.7 26.9 37.5

30.1 64.8 85.9

– 3.3

– – 15.8

2019-03

0.0 255.5 1285.4

0.0 199.8 910.6

0.0 102.4 654.9

19.5 29.0 38.3

26.0 62.6 88.8

– 2.9 –

– – 14.5

2019-04

0.0 265.1 1288.1

0.0 191.0 873.0

0.0 108.6 653.0

23.3 30.6 39.5

30.2 62.0 85.0

– 2.4 –

– – 16.9

2019-05

0.0 240.5 1295.7

0.0 173.5 894.0

0.0 100.8 700.1

23.0 28.9 38.6

37.2 72.4 92.3

– 2.7 –

– – 21.8

2019-06

0.0 246.6 1401.5

0.0 176.1 913.5

0.0 106.8 700.7

22.4 28.3 36.2

49.0 76.9 93.3

– 1.8 –

– – 17.7

2019-07

0.0 223.0 1449.6

0.0 136.3 933.4

0.0 117.5 740.5

22.3 27.4 34.7

48.7 76.9 93.5

– 2.0 –

– – 14.6

2019-08

0.0 210.0 1369.0

0.0 110.1 916.2

0.0 115.4 813.0

22.5 27.3 34.4

51.9 79.0 94.0

– 2.1 –

– – 17.2

2019-09

0.0 190.9 1396.8

0.0 72.0 922.9

0.0 122.9 863.2

21.6 26.5 34.9

45.0 80.7 95.2

– 2.2 –

– – 17.4

www.suntrace.de

Page 87 of 153

4.4 VNSOB

4.4.1 Quality Checks for VNSOB

The results of the quality checks for VNSOB are illustrated in Figure 81. Each horizontal bar contains the relative number of flags identified for a certain parameter. Due to different flag definitions among certain parameter groups, the figure is separated into three parts: radiation parameters, auxiliary parameters and cleaning. From the bar plots in solar measurements, it can be seen that approx. 50% of data is flagged as corrected for most parameter. For DNI and DHI only approx. 35% remain uncorrected. At VNSOB i.e. at Tier 2-stations, both a thermopile pyranometer and and a RSI are installed. Solar irradiance at the RSP is recorded with a Silicium-based (SI) pyranometer, which comprises a similar principle as PV cells. For Tier 1 stations, we mentioned a night-time effect leading to negative values, which must be considered unphysical. For the RSP, no such effect occurs. However, it does for the Thermopile Pyranometer. Quality-Check procedures applied to the dataset flag and correct such negative values, explaining a high percentage in corrected data. Furthermore, GHI is measured based on two principles: a SI-based and a Thermopile Pyranometer. Each is sensitive to a different spectrum. Advanced QCs correct this difference in GHI induced by the sensors sensitivity and, hence, further increasing the percentage of corrected data. Consequently, the observed high percentage of corrected data is not a sign of poor data quality. Instead, the data is, overall, of high quality as there are major data gaps (not counting periods flagged due to cleaning) or unusual values of any solar radiation sensor during daytime. Post-processing concerning the adjustment of SI-based and Thermopile Pyranometer vaues had not been implemented during measurement year 1. Thus, values in the 12-month report show minor deviations from the 24-month report. Corrections have been applied to the entire dataset. For the DHI and DNI measurements of the RSI, 3% of the data is missing. From 1 September 2018 to 18 October 2018, DNI and DHI measurements based on RSI flagged as missing as water got into the instrument-casing. Moreover, some three-component errors in the GHI measure-ments existed. The RSI was replaced on 18 October 2018 during the maintenance visit. For the greater part, reference cells to estimate the impact by soiling have been cleaned suffieciently i.e. according to the agreed cleaning schedule. However December 2017 and May 2018, the cells were cleaned insufficiently. Thus, QCs result in 9% of the time instruments have been maintained insufficiently. During year 2, the cleaning schedule gradually improved, providing overall sufficient cleaning during the second measurement period. Insufficient cleaning will reduce data quality of all parameters. For auxiliary parameters we observe a high data quality as mostly all values are considered as correct. Only some outliers for the temperature measurements of the data logger can be ob-served, leading temperature gradients classified above physical limit. But as the logger tempera-ture is of no other use than for reference, no relevant issue is present due to that.

www.suntrace.de

Page 88 of 153

Figure 81: Results of Quality Checks for VNSOB. Definitions of flags are given in chapter 3.

4.4.2 Seasonal and Diurnal Characteristics for VNSOB

Figure 82 shows seasonal variations in radiation at VNSOB. Monthly DNI averages reveal a strong inter-annual variability. On average, clearest sky conditions were from February through April in 2018 and from January through April in 2019. For both years, GHI exceeded 250 W/m2 in February, March and April. In February 2019, monthly DNI shows exceedingly clear condi-tions leading to a monthly mean above 300 W/m2. DHI shows to have been lower for months with very clear conditions, indicating the presence of less clouds. At VNSOB, December through April were particularly dry with little to no rainfall. This is related to the influence of the monsoon circulation, dividing the annual climate into a dry and a rainy season. In 2017/2018, the dry sea-son even extended from December through September. DNI can be correlated to the monsoon circulation, as high values are to be found at the end of the rainy season. During the rainy sea-son, conditions are more humid. Thus, advection of humid air masses led to a higher cloud-for-mation rate in the region on VNSOB. This shows to have affected insolation, since DHI is high

www.suntrace.de

Page 89 of 153

compared to decreased DNI-values below 150 W/m2. However, the monsoon influence seemed to be more intense in year 2018 when GHI and DNI radiation measurements had lower values from October 2018 to January 2019. In September 2018, water got into the RSI-casing causing an outage. Thus, only GHI values de-rived from the Thermopile Pyranometer are available.

Figure 82: Monthly averaged irradiance measurements for VNSOB. Low DNI and DHI values during September and October 2018 due to RSI-malfunction. Erroenous data has been flagged accordingly and is presented here for completeness. Figure 83 reveals the diurnal characteristics of DNI and GHI at VNSOB. Both DNI and GHI for year 2 lay above year 1 throughout the day. Maximum DNI values are to be found ante meridiem. Throughout the forenoon, DNI values remain relatively constant and gradually decrease as of noon. Furthermore, DNI are clearly right skewed. See therefore the syymetry to the yellow line, which indicates solar noon. The mean daily distribution shows similar characteristics as for very arid regions. Indeed, VNSOB located in the province Binh Tuan, is in one of the most arid regions of Vietnam. At less humid locations, it is often observed that DNI are higher within the first half of the day as cumulus clouds usually form during the afternoon when the ground is at its highest temperatures and rising thermal plumes condensate. This phe-nomenon was often observed for VNSOB and can be examined in more detail in the Monthly Reports.

www.suntrace.de

Page 90 of 153

Figure 83: Average day for the respective year at VNSOB, based on measurements from 1 October 2017 to 30 September 2018 and 1 October 2018 to 30 September 2019, respectively. The yellow line indicates the average time of solar noon.

4.4.3 Soiling Measurements for VNSOB

In this section, the impact of soiling on the PV is assessed by GTI-reference measurements. To follow up of the principles of soiling estimates, please refer to section 4.1.3, where an introduction is given. Analogue, soiling investagations are conducted for VNSOB. The upper plot in Figure 84 shows monthly Global Tilted Irradiance (GTI) over time for RefCell1, RefCell2 and RefCell3. Soiling analysis relies on a well-devined cleaning schedule and station maintenance that complies with the schedule. Station maintenance have been investigated for correctness. Unfortunately, cleaning sheets are only available until June 2019 due to a lack in supply of new sheets. Before then, documents have not been completed sufficiently, e.g. performed cleaning of the Reference Cells has not been documented. The inquiry by our expert Joana Zerbin revealed false cleaning by the station keeper: Unfortunately, both RefCell1 and RefCell2 have been cleaned at every station visit, while RefCell3 has never been cleaned. Consequently RefCell2 will not provide any information on soiling. Only RefCell3 can be applied for correspondance.

www.suntrace.de

Page 91 of 153

Accordingly, Figure 84 reveals an increase in soiling loss from RefCell1 to RefCell3. On monthly averge, soiling losses show to enhanced at RefCell3 throughout most of the observation period. During the dry season (December through April), soiling losses reach up to 2.5% on monthly average. The bottommost plot in Figure 84 shows monthly precipitation at VNSOB and reveals the correlation between rainfall and soiling impact.

Figure 84: Monthly means of GTI measurements at Reference Cells and of monthly soiling loss, alongside monthly precipitation. Figure 85 shows the daily soiling impact for the Reference Cells. For RefCell2, no distinct soiling events become apparent. As previously discussed, cleaning has not been conducted as instructed. This explains why no soiling events are to be seen for RefCell2. Thus soiling rate estimates are based on measurements at RefCell3: As previousely explained, soiling events typically become apparent during dry season. During the dry season of 2018, no events are to be identified as soiling. During the dry season of 2019, however, a major soiling event can be observed. Remaining events are of short (sub-weekly) duration. Therefore, the main soiling event is evaluated as the best estimate for soiling rates at VNSOB and taken for reference. This event suggests a soiling rate of 0.05%. Therefore, see Table 24. Overall, soiling losses may reach -4%. In October 2018, a sudden increase in both soiling loss at RefCell2 and RefCell3 can be observed. This peak is related due to calibration precedures on-site. Therefore, the peak is of no signiificance for data analysis.

www.suntrace.de

Page 92 of 153

Figure 85: Time series of daily soiling impact on reference cells at VNCEH during the two years. Cleaning events by the station keeper are marked by black crosses.Please note: False cleaning at RefCell2

Table 24: Daily soiling rate at VNSOB

Reference Cell 2* Reference Cell 3

Period from to

Days Daily Soiling Rate Period from to

Days Daily Soiling Rate

– – – 2019-01-01 2019-03-10 68 -0.05%

*Due to false cleaning no soiling estimates can be derived

4.4.4 Temperature, Humidity and Precipitation for VNSOB

Climatic conditions of the ambient directly and indirectly influencing PV performance, auxiliary measurements complete extensive assessments on solar resources. Therefore, Figure 86 shows the recorded air temperature, relative humidity and precipitation that have been recorded alongside solar irradiance measurements at VNSOB. For temperature and humidity recordings, shaded areas show the minimum and maximum values of daily averages within one month. Thus, its spread illustrates the inner-monthly variability and shows only little variations in temperature. Seldomly, inner-monthly variations spread by more than 10°C. Such stable conditions are to be expected, since the province Binh Tuan is arid. Monthly temperatures reach approx. 30°C in April and fall down to approx. 27°C during the dry season. The relative humidity shows to stronger seasonal variations. Overall, a correlation between rainfall and relative humidity can be observed: At post-monsoon, humidity falls down to approx. 62%. During the rainy season, humidity approximates 80% on monthly average. Throughout the campaign duration, the last rainy season (2019) shows to have held much more precipitation compared to the first year so that rainfall in year 2 was five time as high as in year 1.

www.suntrace.de

Page 93 of 153

Figure 86: Monthly temperature (upper plot), monthly relative humidity (centre plot) and monthly precipitation (bottom plot).

www.suntrace.de

Page 94 of 153

4.4.5 Wind Measurements for VNSOB

Next to solar measurement instruments, the station at VNSOB is equipped with a wind vane and a cup anemomenter enabling auxiliary measurements. Results of wind measurements are presented as windrose in Figure 87. The windrose reveals winds at VNSOB to rarely exceed 6 m/s. The average wind direction is west-soutwest. 5% of the recorded wind gusts (not shown here) exceed 9.5 m/s.

Figure 87: Windrose for VNSOB.

4.4.6 Measurement Statistics for VNSOB

In this section, measurement statistics are presented in more detail. During the complete two-year measurement period, mean values of radiation parameters are 229 W/m2 for GHI, 181 for DNI W/m2 and 91 W/m2 for DHI, based on Rotating Shadowband Irradiance (RSI) measurements. Table 25 shows mean, minimum and maximum values at VNSOB for both years of the whole-time span related to 1-minute measurements. Precipitation values are based on hourly sums. First-year mean values are calculated for data from 1 October 2017 to 30 September 2018 and second-year mean values from 1 October 2018 to 30 September 2019, respectively.

www.suntrace.de

Page 95 of 153

Table 25: Statistics on the whole dataset providing mean, minimum and maximum values of 1-minute values calculated from daily means due to availability

PARAMETER UNIT YEAR 1 YEAR 2

MEAN MIN MAX MEAN MIN MAX GHI (TH.PYR) W/m2 225 0 1479 236 0 1417 GHI (RSI) W/m2 223 0 1603 236 0 1519 DNI (RSI) W/m2 168 0 1025 193 0 991 DHI (RSI) W/m2 93 0 792 90 0 857 GTI (CLEANED OFTEN) W/m2 228 0 1514 238 0 1428 GTI (CLEANED RARELY) W/m2 228 0 1514 238 0 1429 GTI (NEVER CLEANED) W/m2 226 0 1501 234 0 1403 TEMPERATURE °C 30.6 17.7 44.0 31.0 19.8 45.8 HUMIDITY % 73 30 100 73 35 96 PRECIPITATION (ANNUAL ACCUM.) mm 121.6 610.0

AIR PRESSURE hPa 1002 970 1010 1002 971 1011 WIND SPEED m/s 2.3 0.0 14.7 2.1 0.0 16.8 WIND GUSTS m/s – – 19.5 – – 20.4 WIND DIRECTION °N 171.1 – – 169.1 – –

At VNSOB, year 2 reveals a significant increase in GHI and DNI. Indicating strong intra-annual variability, GHI increased by approx. 6% from year 1 to year 2. DNI increased by even 18 %. Furthermore, year 2 was characterised by five times as much precipitation compared to year 1. Thus, during year 2 clouds are to be more present. However, DHI did not significantly increase. Figure 86 suggests relative humidity to have been amplified during year 1, which is likely to lead to an increase in scattered light in the atmosphere. Thus, annual means as listed above reveal no distinct intra-annual increase in humidity. Temperatures and air pressure show to have been constant throughout the campaign duration. This reflects the overall stable inter-annual conditions, as stated before. On annual average, soiling becomes visible at Reference Cell 3 i.e. when never cleaning the module. Reference Cell 2, however, does not show a reduction in the annual average yield. This is to be explained by false cleaning: As the station keeper cleaning both RefCell1 and RefCell2, no soiling impact i.e. no difference between GTI (cleaned often) and GTI (cleaned rarely) is to be expected.

www.suntrace.de

Page 96 of 153

As minimum and maximum values are less conservative and can be partly affected by measurement errors, Table 26 shows the 5th and 95th percentile for the main parameters. As irradiation during night is 0 W/m2, only daytime values have been considered for the radiation parameters.

Table 26: Percentiles of measurement paramters for the whole two-year measurement period

PARAMETER UNIT 5-PERCENTILE 95-PERCENTILE GHI (TH.PYR) W/m2 16 987 GHI (RSI) W/m2 17 984 DNI (RSI) W/m2 0 873 DHI (RSI) W/m2 15 465 TEMPERATURE °C 25.6 40.4 RELATIVE HUMIDITY % 45.0 90.4 WIND GUSTS m/s 0.8 9.4

Even though highest values for GHI exceeded 1600 W/m2 during the measurement period, less than 5 % of GHI based on RSI-measurements exceeded 984 W/m2. For Thermopile Pyranometer GHI-measurements the value increases to 987 W/m2, since the instruments records a broader spectrum. At the lower part, 95 % of the data lies above 16 W/m2 for GHI and 15 W/m2 for DHI. For further details, Table 26 and Table 27 provide monthly averages for both years of the measurement data for the most important parameters regarding PV operation. Please note for Table 26: In September 2018, water in the casing caused an RSI-outage. Thus, DNI and DHI values for this month are not representative, but still listed for completeness.

www.suntrace.de

Page 97 of 153

Table 27: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 1 *Values not representative, due to RSI ouatge

MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

PRECIP [MM] – – SUM

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2017-10

0.0 173.9 1291.8

0.0 102.6 929.3

0.0 99.3 734.4

20.9 26.6 34.4

43.1 86.1 100.0

– – 79.3

– 1.3 –

– – 19.5

2017-11

0.0 179.0 1361.7

0.0 127.6 911.9

0.0 97.8 733.2

21.2 26.8 35.21

47.8 81.8 99.1

– – 25.6

– 1.5 –

– – 13.4

2017-12

0.0 190.9 1294.8

0.0 167.1 937.2

0.0 87.3 698.2

18.6 26.5 35.1

33.1 69.7 96.9

– – 0.8

– 1.8 –

– – 16.3

2018-01

0.0 203.4 1221.6

0.0 165.8 958.8

0.0 95.5 701.2

18.9 26.9 34.6

33.2 69.4 96.4

– – 0.2

– 1.7 –

– – 12.3

2018-02

0.0 255.2 1407.5

0.0 258.9 940.6

0.0 76.2 737.5

16.2 26.3 35.1

30.9 62.2 93.1

– – 0.0

– 2.0 –

– – 13.4

2018-03

0.0 274.4 1252.1

0.0 250.4 941.2

0.0 87.1 693.5

21.1 27.8 35.4

30.2 65.7 94.5

– – 4.2

– 1.9 –

– – 12.7

2018-04

0.0 286.6 1311.8

0.0 264.3 1026.0

0.0 83.6 607.8

20.6 29.0 37.2

31.9 65.0 90.7

– – 0.0

– 2.1 –

– – 13.3

2018-05

0.0 244.2 1325.9

0.0 200.0 921.8

0.0 90.3 676.1

24.2 29.6 37.3

39.2 69.6 93.2

– – 5.4

– 1.8 –

– – 13.6

2018-06

0.0 214.7 1467.5

0.0 141.8 1018.8

0.0 106.9 705.8

23.5 28.5 36.4

41.7 73.4 95.3

– – 5.8

– 3.0 –

– – 17.6

2018-07

0.0 230.3 1326.0

0.0 142.2 958.5

0.0 120.6 792.4

23.3 28.3 37.6

39.8 74.8 94.1

– – 0.1

– 4.3 –

– – 17.8

2018-08

0.0 213.6 1467.3

0.0 117.3 945.3

0.0 124.3 704.7

23.6 28.1 36.7

43.6 73.5 93.5

– – 0.1

– 4.3 –

– – 19.0

2018-09

0.0 236.1 1479.1

0.0* 4.8* 774.2

0.0* 0.9* 232.1

21.7 27.2 36.1

45.1 79.9 95.1

– – 0.1

– 2.0 –

– – 19.4

www.suntrace.de

Page 98 of 153

Table 28: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 2 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [degC] MIN MEAN MAX

RH [%] MIN MEAN MAX

PRECIP [MM] MIN MEAN MAX

WSPEED [M/S] MIN MEAN MAX

WGUST [M/S] MIN MEAN MAX

2018-10

0.0 231.8 1212.0

0.0 123.7 934.6

0.0 43.1 713.5

19.5 27.4 35.5

38.9 76.2 95.6

– – 50.3

– 1.4 –

– – 10.7

2018-11

0.0 196.4 1206.7

0.0 180.5 946.3

0.0 78.7 709.6

20.9 26.9 35.0

41.2 77.3 95.4

– – 142.6

– 1.5 –

– – 11.1

2018-12

0.0 180.4 1238.1

0.0 136.4 933.4

0.0 95.5 678.7

21.5 27.1 35.0

46.5 74.2 94.3

– – 3.9

– 1.6 –

– – 14.2

2019-01

0.0 218.3 1325.1

0.0 222.8 977.8

0.0 76.5 675.0

18.1 26.7 34.9

34.9 63.7 93.7

– – 1.0

– 2.0 –

– – 13.6

2019-02

0.0 270.1 1130.0

0.0 304.2 991.0

0.0 61.4 634.8

19.4 26.6 34.2

34.6 66.8 91.9

– – 0.0

– 2.0 –

– – 12.9

2019-03

0.0 268.2 1413.1

0.0 227.1 956.2

0.0 98.0 856.5

18.4 27.9 36.0

38.6 68.3 89.9

– – 0.4

– 1.9 –

– – 13.1

2019-04

0.0 273.3 1256.1

0.0 235.8 949.6

0.0 91.0 708.5

20.3 29.6 36.7

38.5 66.9 87.5

– – 0.1

– 1.9 –

– – 12.2

2019-05

0.0 247.4 1202.9

0.0 199.5 916.1

0.0 90.0 699.4

23.5 29.3 38.5

44.7 74.9 93.6

– – 29.5

– 1.8 –

– – 17.0

2019-06

0.0 250.8 1405.8

0.0 203.1 911.7

0.0 97.8 706.8

24.2 29.3 38.3

44.2 74.6 93.8

– – 55.0

– 2.6 –

– – 16.7

2019-07

0.0 237.3 1363.8

0.0 163.9 935.6

0.0 109.9 716.4

22.5 28.1 37.1

47.1 78.0 94.3

– – 139.8

– 2.7 –

– – 18.3

2019-08

0.0 237.2 1416.6

0.0 171.6 923.4

0.0 106.4 855.4

23.3 28.2 37.2

44.8 77.6 94.5

– – 76.4

– 3.5 –

– – 20.4

2019-09

0.0 219.8 1393.6

0.0 137.5 945.0

0.0 115.8 784.4

22.5 27.4 36.4

44.4 80.1 95.2

– – 111.0

– 2.6 –

– – 18.6

www.suntrace.de

Page 99 of 153

4.5 VNTRA

4.5.1 Quality Checks for VNTRA

Figure 88 illustrates the results of performed QC on the data recorded at VNTRA. Each horizon-tal bar contains the relative amount of flags identified for a certain parameter. Due to the differ-ent flag definitions among certain parameter groups, the figure is separated into three parts: Ra-diation parameters, auxiliary parameters and station maintenance i.e. cleaning of instruments. For radiation parameters, it can be seen that approx. 50% of the measurements are marked as corrected. For DNI measurements however, approx. 90% of the data remain uncorrected by the pre-defined QC, indicating the high level of Thermopile Pyrheliometer measurements. Radiation measurements performed by thermopile-based instruments often show negative val-ues during the night. This behaviour is a consequence of the instrument as it emits more radia-tion (infra-red) than it receives because it is generally warmer than effective environmental tem-perature at night. It leads to values mostly in the range of –1 to –4 W/m2. This effect cannot be observed in solar power plants, e. g. photovoltaic, so the values are set to 0 W/m2. Thus, a high percentage in corrected data is no sign of poor data quality. For the auxiliary measurements, the dataset is of very high quality since according to the flags all the data is classified as correct after Quality Check. One essential role to guarantee a high measurement data quality is to keep the station clean to avoid disturbing effects from the environment influencing the measure-ments. At VNTRA, to approx. 95% of the campaign duration the station has been maintained sufficiently.

www.suntrace.de

Page 100 of 153

Figure 88: Results of Quality Checks for VNTRA. Definitions of flags are given in chapter 3.

4.5.2 Seasonal and Diurnal Characteristics for VNTRA

Figure 89 shows the seasonal variations in radiation at VNTRA. Compared to previously pre-sented results for the stations VNHAN and VNDAN in Northern Vietnam, the data recorded at VNTRA reveals a lower inner-annual variability with regard to irradiance. Due to its geographical proximity to VNCEH and VNSOB in Southern Vietnam, it is not surprising that both stations show similar monthly patterns. Furthermore, the magnitude of irradiance at VNTRA corresponds to values recorded at VNCEH and VNSOB. From June through August 2018, stations VNTRA and VNCEH have recorded high monthly DHI that exceed monthly DNI, showing high presence in clouds. In 2019, this observation can only be found for VNTRA. From February to May 2018 and 2019, VNTRA shows to have been less sunny than VNSOB, as can be seen in the DNI val-ues.

www.suntrace.de

Page 101 of 153

Monsoon-circulation induced Southerly winds lead to increasing atmospheric humidity and cloud formation and, accordingly, to a decrease in GHI and DNI. In contrary, DHI increased and exceeded 100 W/m2 from February through August in both years. Clearest periods are to be found near the turn of the year, when monthly averaged DNI exceed 200 W/m2. Throughout the observation period, January 2019 shows to have been the clearest month, enabling DNI of al-most 250 W/m2. Accordingly, maximum GHI values are to be found from January through April with values above 250 W/m2.

Figure 89: Monthly averaged irradiance measurements from VNTRA. Figure 90 shows the diurnal characteristics of irradiance at VNTRA. For a location with no cloud coverage, a perfect symmetry to solar noon is to be expected. During the measurement cam-paign many days with almost clear sky conditions were found. Thus, the high level of symmetry is evidence for predominantly clear conditions at VNTRA. At noon, annual mean DNI values touch 500 W/m2. Nonetheless, a slight skewness can be observed. As discussed earlier, in hot and dry locations it is often observed that DNI is higher within the first half of the day as cumu-lus clouds usually form during the afternoon and rising thermal plumes condensate. During late afternoon increasing cloudiness formed by convection cause DNI to decrease. This behaviour induces the visible right skewedness. Diurnal GHI values touch 800 W/m2.

www.suntrace.de

Page 102 of 153

Figure 90: Average day of the year for VNTRA. The yellow line shows the average time of solar noon.

4.5.3 Soiling Measurements for VNTRA

In this section, the impact of soiling on the PV is assessed by GTI-reference measurements. To follow up of the principles of soiling estimates, please refer to section 4.1.3, where an introduction is given. Analogue, soiling investagations are conducted for VNTRA. At VNTRA, data analysis gave raise to suspect false cleaning: Station keepers are instructued to stricly follow the pre-defined cleaning schedule. Since initial steps on behalf of soiling analysis did not show the expected behaviour, extended investigations suggested an inverted cleaning order. Thus, we assume RefCell3 to have been cleaned regularly i.e. at each station visit and RefCell1 having never been cleaned. Verification of this finding is underway and will be clarified. However, an inverted cleaning order shows most plausible results. Therefore, for VNTRA wie will assume RefCell1 = RefCell3 and RefCell3 = RefCell1 hereinafter until further verification. The upper plot in Figure 91 shows monthly Global Tilted Irradiance (GTI) over time for RefCell1, RefCell2 and RefCell3. As previously discussed, precipitation often interferes soiling analysis, as it “cleans” all three Reference Cells simultaneously. Thus, in accordance to the dry season, an increase in soiling loss can be observed from RefCell2 to RefCell3 from October through April for both years. On monthly average, soiling losses may cause as much as 1.2% less energy uptake.

www.suntrace.de

Page 103 of 153

Figure 91: Monthly means of GTI measurements at Reference Cells and of monthly soiling loss. Under the assumption of inverted cleaning order, Figure 92 shows distinct soiling impact at both RefCell2 and RefCell3. As previousely explained, soiling events typically become apparent during dry season. Since no precipitation measurements were conducted on-site, an interpretation of the timeseries is less precise. Following up on rainfall in Southern Vietnam, soiling events are expected to become apparent from January through April of each year. Indeed, during this period soiling events can be observed. At RefCell3, daily soiling losses reach down to 2%.

Figure 92: Time series of daily soiling impact on reference cells at VNTRA during the two years. Cleaning events by the station keeper are marked by black crosses.

www.suntrace.de

Page 104 of 153

The methodology of periods that show clear soiling is presented in section 4.1.3 and will not be discussed at this point. Selected periods for soiling rate estimation at VNTRA are given in Table 29 alongside the found daily soiling rate. Please note that chosen periods are considerably short due to frequent rainfall in this region.

Table 29: Daily soiling rate at VNTRA

Reference Cell 2 Reference Cell 3

Period from to

Days Daily Soiling Rate Period from to

Days Daily Soiling Rate

2018-02-02 2018-02-11 9 -0.10% 2018-01-20

2018-02-11 22 -0.07%

2018-03-02 2018-03-11 18 -0.05% 2018-02-15

2018-03-10 23 -0.04%

2019-03-02 2019-03-20 22 -0.10% 2019-03-02

2019-03-20 18 -0.10%

Due to cleaning of RefCell2, considered timespans are on average shorter for RefCell2 than fro RefCell3. Furthermore, Table 29 reveals a saturation effect: On average, shorter timespans show higher daily soiling rates compared to longer periods. This is due to the contamination of the cell gradually reaching saturation, hence, the soiling rate is reduced once the cell is already dirty. However, this must not be considered as self-cleaning, since the degree of contamination i.e. losses in energy uptake will remain. Averaging the periods given yields the following mean daily soiling rates:

• RefCell 2, mean Daily Soiling Rate: -0.08% • RefCell 3, mean Daily Soiling Rate: -0.07%

This translates to an average daily net loss of 0.08% in solar irradiance uptake. Thus, soiling shows to have timewise significant impact on PV performance. Since saturation effects affect measurements at RefCell3, it is not surprising to observe a lower soiling rate compared to RefCell 2. This, however, means that soiling rates obtained from RefCell2 underestimate actual soiling.

www.suntrace.de

Page 105 of 153

4.5.4 Temperature and Humidity for VNTRA

Climatic conditions of the ambient directly and indirectly influencing PV performance, auxiliary measurements complete extensive assessments on solar resources. Therefore, Figure 93 shows the recorded air temperature and relative humidity that have been recorded alongside solar irradiance measurements at VNTRA. For temperature and humidity recordings, shaded areas show the minimum and maximum values of daily averages within one month. Thus, its spread illustrates the inner-monthly variability and shows only little variations in temperature. Seldomly, inner-monthly variations spread by more than 10°C. Monthly temperatures reach approx. 29°C in April and fall down to to approx 25°C during the dry season. The relative humidity shows stronger seasonal variations. Overall, a correlation between rainfall and relative humidity can be observed: At post-monsoon, humidity falls down to approx. 70%. During the rainy season, humidity approximates 87% on monthly average. In conclusion, VNTRA show to slightly colder, yet reveals slightly higher relative humidity than VNSOB.

www.suntrace.de

Page 106 of 153

Figure 93: Monthly temperature and relative humidity at VNTRA

www.suntrace.de

Page 107 of 153

4.5.5 Wind Measurements for VNTRA

Next to solar measurement instruments, the station at VNTRA is equipped with a wind vane and a cup anemomenter enabling auxiliary measurements. Results of wind measurements are presented as windrose in Figure 94. Here, the extend of the windrose illustrates the percentage to which given wind speed and direction are found among the data. The windrose reveals winds at VNTRA to rarely exceed 4 m/s. The average wind direction is south-southeast to south. 5% of the recorded wind gusts (not shown here) exceed 7.9 m/s.

Figure 94: Windrose for VNTRA.

4.5.6 Measurement Statistics for VNTRA

In this section, measurement statistics are presented in more detail. During the complete two-year measurement period, mean values of radiation parameters are 221 W/m2 for GHI, 158 W/m2 for DNI and 101 W/m2 for DHI, based on thermopile pyranometer measurements. Table 30 shows mean, minimum and maximum values at VNTRA for both years of the whole-time span related to 1-minute measurements.

www.suntrace.de

Page 108 of 153

Table 30: Statistics on the whole dataset providing mean, minimum and maximum values of 1-minute values calculated from daily means due to availability

PARAMETER UNIT YEAR 1 YEAR 2

MEAN MIN MAX MEAN MIN MAX GHI (TH.PYR) W/m2 217 0 1453 221 0 1384 DNI (TH.PYR) W/m2 147 0 978 162 0 980 DHI (TH.PYR) W/m2 105 0 1054 100 0 759 GTI (CLEANED OFTEN) W/m2 218 0 1469 221 0 1439 GTI (CLEANED RARELY) W/m2 217 0 1463 221 0 1442 GTI (NEVER CLEANED) W/m2 218 0 1470 221 0 1438 TEMPERATURE °C 34.6 24.5 50.8 35.3 25.9 51.0 HUMIDITY % 81 32 100 80 26 98 AIR PRESSURE hPa 1001 992 1009 1001 992 1009 WIND SPEED ms-1 2.3 0.0 17.7 2.4 0.0 20.2 WIND GUSTS ms-1 – – 24.7 – – 25.1 WIND DIRECTION °N 178.0 – – 162.2 – –

In correspondence to VNSOB, irradiance measurements at VNTRA show higher annual mean GHI and DNI for year 2 compared to year 1. Yet, the intra-annual increase is less than at VNSOB. This reveals clearer conditions for year 2 at both stations. Furthermore, irradiance values are approx. 2% lower at VNTRA compared to VNSOB. Annual DHI shows to have decreased from year 1 to year 2 by 5%. Annual means as listed above reveal no distinct intra-annual increase in humidity. Year two shows to have been slightly warmer than year 1. An increase in temperature by 0.7°C is reported. On annual average, soiling does not become apparent. As minimum and maximum values are less conservative and can be partly affected by measurement errors, Table 31 shows the 5th and 95th percentile for the main parameters. As irradiation during night is 0 W/m2, only daytime values have been considered for the radiation parameters.

www.suntrace.de

Page 109 of 153

Table 31: Percentiles of measurement paramters for the whole two-year measurement period

PARAMETER UNIT 5-PERCENTILE 95-PERCENTILE GHI (TH.PYR) W/m2 15 974 DNI (TH.PYR) W/m2 0 845 DHI (TH.PYR) W/m2 14 461 TEMPERATURE °C 30.5 45.8 RELATIVE HUMIDITY % 51 94 WIND GUSTS ms-1 1.1 7.9

Even though highest values of GHI exceeded 1450 W/m2 during the measurement period, less than 5% exceeded 974.2 W/m2 during daytime. The fact that only 5% of the data shows relative humidity values below 51 % highlight that the site underlies very humid conditions. Moreover, temperature did not drop below 30.5°C in 95 % of the measurement period. For further details, Table 32 and Table 33 provide monthly averages for both years of the measurement data for the most important parameters regarding PV operation.

www.suntrace.de

Page 110 of 153

Table 32: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 1 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [degC] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] MIN MEAN MAX

WGUST [M/S] MIN MEAN MAX

2017-10

0.0 196.1 1378.0

0.0 118.0 935.3

0.0 104.0 723.8

23.4 26.8 34.6

45.5 85.3 100.0

– 2.1 –

– – 23.9

2017-11

0.0 192.1 1296.3

0.0 145.8 935.0

0.0 98.5 647.6

21.8 26.8 33.3

48.7 83.5 100.0

– 2.0 –

– – 16.5

2017-12

0.0 202.1 1161.5

0.0 177.1 945.7

0.0 85.0 637.1

19.7 26.0 33.0

42.9 76.5 100.0

– 2.4 –

– – 16.3

2018-01

0.0 203.0 1222.7

0.0 158.4 977.7

0.0 93.5 620.5

20.7 26.5 33.8

45.6 78.5 100.0

– 2.0 –

– – 12.7

2018-02

0.0 250.6 1206.1

0.0 223.5 954.1

0.0 88.2 640.0

18.9 26.3 34.5

35.0 70.3 97.4

– 2.2 –

– – 10.2

2018-03

0.0 256.2 1216.6

0.0 195.2 948.5

0.0 104.8 669.2

22.5 27.9 36.9

34.8 71.6 100.0

– 2.2 –

– – 11.8

2018-04

0.0 261.4 1319.9

0.0 182.0 920.6

0.0 113.1 714.4

23.5 28.8 37.5

32.3 72.0 98.1

– 2.3 –

– – 12.8

2018-05

0.0 247.2 1384.3

0.0 180.5 903.4

0.0 98.3 652.4

23.2 28.0 35.5

41.3 82.6 100.0

– 2.1 –

– – 18.5

2018-06

0.0 205.2 1297.7

0.0 93.9 879.8

0.0 127.7 736.9

23.0 27.3 34.4

50.3 86.0 100.0

– 2.4 –

– – 21.9

2018-07

0.0 187.5 1315.0

0.0 82.7 883.0

0.0 119.5 705.1

22.7 27.1 34.0

52.4 87.5 100.0

– 2.9 –

– – 19.6

2018-08

0.0 193.1 1361.5

0.0 79.8 870.8

0.0 127.2 1054.0

22.7 27.0 33.2

56.7 86.9 100.0

– 2.8

– – 24.7

2018-09

0.0 211.4 1452.6

0.0 129.0 906.6

0.0 106.9 734.8

22.1 26.9 33.8

56.1 87.3 100.0

– 2.0 –

– – 14.9

www.suntrace.de

Page 111 of 153

Table 33: Monthly Statistics on Radiation Data in W/m2 and some auxiliary data for measurement year 2 MONTH

GHI [W/M2] (th. Pyr) MIN MEAN MAX

DNI [W/M2] (RSI) MIN MEAN MAX

DHI [W/M2] (RSI) MIN MEAN MAX

TEMP [°C] MIN MEAN MAX

RH [%] MIN MEAN MAX

WSPEED [M/S] – MEAN –

WGUST [M/S] – – MAX

2018-10

0.0 248.1 1384.3

0.0 209.9 922.9

0.0 88.5 680.4

22.6 27.6 33.9

41.2 81.7 98.4

– 2.0 –

– – 15.3

2018-11

0.0 196.3 1166.7

0.0 167.1 951.7

0.0 80.9 605.1

23.4 27.4 34.0

48.0 81.9 97.9

– 2.3 –

– – 17.2

2018-12

0.0 194.0 1215.3

0.0 148.3 939.1

0.0 95.3 631.3

22.9 27.4 33.7

41.0 79.5 97.7

– 2.1 –

– – 14.8

2019-01

0.0 220.1 1129.1

0.0 219.6 979.7

0.0 73.6 623.1

20.1 26.9 33.4

31.5 71.9 95.8

– 2.2 –

– – 14.4

2019-02

0.0 253.3 1097.9

0.0 246.2 964.4

0.0 75.8 640.1

21.5 27.6 34.6

27.3 69.4 95.2

– 2.5

– – 11.4

2019-03

0.0 257.1 1293.7

0.0 195.0 931.5

0.0 107.4 692.9

22.0 28.4 36.3

25.6 72.4 93.5

– 2.4 –

– – 19.2

2019-04

0.0 258.8 1206.8

0.0 189.6 950.5

0.0 106.8 658.4

22.0 29.4 38.1

31.6 75.9 95.6

– 2.4 –

– – 25.1

2019-05

0.0 227.5 1275.9

0.0 139.0 868.7

0.0 115.4 702.0

22.5 28.6 36.7

48.1 82.6 97.7

– 2.2 –

– – 20.9

2019-06

0.0 221.8 1356.4

0.0 140.2 902.9

0.0 109.1 659.3

23.1 28.2 35.7

42.4 84.0 96.9

– 2.4 –

– – 18.7

2019-07

0.0 200.9 1302.0

0.0 99.5 907.7

0.0 119.7 708.2

22.6 27.3 34.0

46.3 85.0 96.6

– 2.4 –

– – 24.3

2019-08

0.0 189.7 1297.0

0.0 93.1 876.2

0.0 113.7 728.2

23.4 27.3 33.5

61.4 86.1 97.1

– 2.7 –

– – 19.2

2019-09

0.0 190.8 1292.2

0.0 98.1 887.3

0.0 112.6 759.2

22.7 26.9 34.6

47.6 86.5 96.2

– 2.4 –

– – 20.5

www.suntrace.de

Page 112 of 153

4.6 Comparison of Monthly Averages

In order to provide a quick access to seasonal variations in irradiance at all five stations, Figure 95 and Figure 96 illustrate monthly DNI and GHI values, respectively. Detailed analysis on irradiance and auxiliary measurements was given in previous sections. This section is inteded to compare stations. As mentioned before, RSI-outage at VNSOB in September 2018 lead to missing DNI and DHI values. Accordingly, values have been taken out in the figure.

Figure 95: Biannual DNI measurement comparison between the Helioscale stations in Vietnam. Monthly average values are plotted; annual means are listed accordingly. Both, Figure 95 and Figure 96 show highest DNI and GHI values to have been recorded at VNSOB throughout the campaign. Eventhough VNCEH is by far located at highest altitude among the station network, it does not comprise highest values. Instead, a geographical trend can be observed: The Northern Stations VNDAN and VNCEH reveal much lower annual GHI and DHI. Especially, DNI shows to significantly decrease, the further north the station is located.

www.suntrace.de

Page 113 of 153

Rainy season and dry season are temporarily not the same throughout Vietnam. Thus, a seasonal shift can be observed among the stations: Northern stations show highest irradiance values from May through September, whereas Southern statoins reveal their maxima from January through August. This is in correlation with the temporal shift of the monsoon season.

Figure 96: Biannual GHI measurement comparison between the Helioscale stations in Vietnam. Monthly average values are plotted; annual means are listed accordingly.

www.suntrace.de

Page 114 of 153

5 Estimation of long-term best estimate (P50) for GHI & DNI and corresponding uncertainties

On-site measurements recorded irradiance alongside auxiliary parameters for more than two years. This data is used for long-term estamation of GHI and DNI as presented in this chapter.

5.1 Methodology

For evaluating the long-term average of GHI and DNI at the five locations, two different solar resource data sets for each location have been analysed: site specific long-term satellite-derived DNI and GHI data obtained from Solargis (SG), as well as the on-site measurements taken at the site for more than two complete years.

Since October 2017, 26 months of high-precision on-site measurements are available. It is supposed that these on-site measurements from these locations already had been used to improve the model for deriving the GHI and DNI by satellite data. However an additional improvement was possible by site-specific adaption of the long-term satellite derived GHI and DNI data set by on-site measurements for the five locations. For this, the on-site measurements have been inter-compared to the long-term satellite derived GHI and DNI data from SG for the same period to determine corresponding correction functions. The found correction functions have been used to adjust the SG-dataset for the whole period from January 1999 up to November 2019.

The best estimate for the long-term annual GHI and DNI at the five sites are derived from the adjusted long-term SG datasets, following the recommended procedure of the International Electrotechnical Commission (IEC), as documented in (IEC TC117 2017, 62862-1-2 ).

The total long-term uncertainty of the long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainties of GHI measured by the Secondary Standard Pyranometer are assumed to be 1.5% for all five sites. The long-term instrument uncertainties for the DNI measured with the First Class pyrheliometer installed at VNHAN, VNCEH and VNTRA are assumed to be 2.0% and for the DNI measured with the Rotating Shadowband Pyranometer installed at VNDAN and VNSOB are assumed to be 2.5% respectively. The uncertainty due to calibration and maintenance issues is assumed to be 0.5 %. Thus, the total long-term uncertainties of measured GHI and DNI data are obtained by considering these two factors according to the Gaussian error law of propagation and found to be about 1.6 % for the GHI measured by the Secondary Standard Pyranometer, 2.1 % for the DNI measured with the First Class pyrheliometer and 2.5 % for the DNI measured with the Rotating Shadowband Pyranometer.

The uncertainty of satellite data is reduced by using the adjustment methodology. According to (Mieslinger, et al. 2014), the uncertainty of the adaption methodology is approximately 3 % on average, which varies depending on the site, the data source, and the duration of the overlapping period with on-site measurements. The overlapping period with on-site measurements is 26 months. Thus, for this study an uncertainty of 2 % is assumed for the adaption methodology for GHI and 3 % for DNI since GHI showed better correlation in comparison to DNI.

www.suntrace.de

Page 115 of 153

Following (Meyer, et al. 2009), the uncertainty due to considering only a few years reduces with the number of years. In this analysis, the main base for fixing the long-term value is more than 20 years dataset of adapted SG. Thus, the contribution to uncertainty due to inter-annual variations is reduced by the factor 1/Ö20, which is around 0.224. This is taken as input to derive the overall uncertainty as shown in detail in the corresponding tables below.

5.2 VNHAN

5.2.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data

Figure 97 shows a scatter plot of original and adjusted SG GHI and DNI values against the on-site measurements in hourly time resolution for the overlapping period from 1 October 2017 up to 30 November 2019.

During this overlapping period, the GHI on-site measurement data had an average of 150 W/m2, while satellite-derived SG data had an average of 152 W/m2. Thus, the original SG dataset overestimates in total 1.3 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG GHI data by on-site measurements, the average of the adjusted GHI data amounts also to 150 W/m2. The DNI on-site measurement data had an average of 73 W/m2 during the overlapping period, while satellite-derived SG data had an average of 72 W/m2. Thus, the original SG data set underestimates in total 1.4 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG DNI data by on-site measurements the average of the adjusted DNI data stays at 72 W/m2.

Figure 97: Scatter plots of GHI (left) and DNI (right) displaying on-site-measured hourly data points against original and adjusted GHI and DNI from SG. The black line represents the 1:1 relationship.

www.suntrace.de

Page 116 of 153

The frequency distribution of adjusted SG satellite-derived GHI and DNI matches very well to ground-measured values as can be seen in Figure 98. Figure 99 shows the Monthly averages of GHI and DNI for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 98: Frequency distribution of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 99: Monthly averages of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

5.2.2 Determination of long-term average

The long-term best estimate for GHI at this site amounts to 148 W/m2, which is equivalent to 1295 kWh/m2 per year or 3.6 kWh/m2 per day.

The long-term best estimate for DNI at this site amounts to 74 W/m2, which is equivalent to 650 kWh/m2 per year or 1.8 kWh/m2 per day.

www.suntrace.de

Page 117 of 153

5.2.2.1 GHI

For site VNHAN, six different input data sets have been assessed, from which finally the one from SG adjusted by on-site measurements was taken into account for determining the long-term annual best estimate of GHI. The main results can be found in Table 34 and Figure 100, which visualises the GHI long-term averages for these eight datasets together with the best estimate (green line).

Table 34: Overview of long-term averages from various DNI datasets.

Figure 100: Overview of the long-term GHI values from different data sets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 118 of 153

1.1.1.1 DNI

For site VNHAN, five different input datasets have been assessed, from which the SG-dataset (adjusted by on-site measurements) was taken into account for determining the long-term annual best estimate of GHI. Main results can be found in Table 35 and Figure 101, which visualises the GHI long-term averages for these five datasets together with the best estimate (green line).

Table 35: Overview of long-term averages from various DNI data sets.

Figure 101: Overview of the long-term DNI values from different datasets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 119 of 153

5.2.3 Analysis of uncertainty

As mentioned in chapter 5.1 Methodology, the total long-term uncertainty of long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainty of GHI measured by the Secondary Standard Pyranometer is assumed to be 1.6 % including the uncertainty due to calibration and maintenance. The long-term instrument uncertainty of DNI measured by the First Class Pyrheliometer is assumed to be 2.1 % including the uncertainty due to calibration and maintenance. For this study the uncertainty of the adaption methodology is assumed to be 2 % for GHI and 3 % for DNI. Based on the adapted SG data set, the inter-annual variability of GHI is 10 W/m2, approximately equivalent to ± 6.9 %. Compared to other regions in the world, this is a quite high volatility for GHI. The inter-annual variability of DNI is with 11 W/m2, approximately equivalent to ± 15.0 %, very high.

Based on the above-mentioned values, the total uncertainty of the long-term best estimates of GHI (P50) is found to be 3.2 % associated with the inter-annual variability related to multiple years and 7.6 % related to a single year. The total uncertainty of the long-term best estimates of DNI (P50) is found to be 5.6 % associated with the inter-annual variability related to multiple years and 16.0 % related to a single year. Further details can be found in Table 36.

Table 36: Explanation of derivation of the total long-term uncertainty of GHI and DNI following Meyer et

al. (2008).

Calculation of long-term uncertainty of long-term average of GHI and DNI for multiple year and single year

overall uncertainty adjustment uncertainty

uncertainty due to temporal coverage on-site meas.

multiple year

single year

multiple year single year A B Cmy=

Csy/√n Csy Umy=√(A2+B2+Cmy

2) Usy=√(Umy

2+Cs

y2)

abs. [W/m2]

rel. [%]

abs. [W/m2]

rel. [%] [%] [%] [%] [%]

best estimate (P50) GHI

5 3.2 11 7.6 2 1.6 1.5 6.9

best estimate (P50) DNI

4 5.6 12 16.0 3 2.1 3.4 15.0

www.suntrace.de

Page 120 of 153

5.2.4 Annual cycle of GHI and DNI

Table 37 and Figure 102 for GHI and Table 38 and Figure 103 for DNI, respectively, show the long-term annual cycle of the best estimate (SGadjusted) for GHI and DNI in comparison to the original SG dataset, as well as the datasets from CMSAF-SARAH, 3Tier (GHI), DLR-ISIS, NASA-SSE and MN7.

Table 37: Long-term monthly average values of GHI.

Figure 102: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 121 of 153

Table 38: Long-term monthly average values of DNI.

Figure 103: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 122 of 153

5.3 VNDAN

5.3.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data

Figure 104 shows a scatter plot of original and adjusted SG GHI and DNI values against the on-site measurements in hourly time resolution for the overlapping period from 1 October 2017 up to 30 November 2019.

During this overlapping period, the GHI on-site measurement data had an average of 189 W/m2, while satellite-derived SG data had an average of 197 W/m2. Thus, the original SG dataset overestimates in total 4.2 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG GHI data by on-site measurements, the average of the adjusted GHI data amounts also to 189 W/m2. The DNI on-site measurement data had an average of 125 W/m2 during the overlapping period, while satellite-derived SG data had an average of 143 W/m2. Thus, the original SG data set overestimates in total 14.4 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG DNI data by on-site measurements the average of the adjusted DNI data amounts to 126 W/m2.

The frequency distribution of adjusted SG satellite-derived GHI matches very well to ground-measured values, for DNI not, as can be seen in Figure 105. However, Figure 106 shows that the monthly averages of both adjusted GHI and adjusted DNI satellite data fit better to the on-site measurements in comparison to the original satellite data for the overlapping period.

Figure 104: Scatter plots of GHI (left) and DNI (right) displaying on-site-measured hourly data points against original and adjusted GHI and DNI from SG. The black line represents the 1:1 relationship.

www.suntrace.de

Page 123 of 153

Figure 105: Frequency distribution of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 106: Monthly averages of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

5.3.2 Determination of long-term average

The long-term best estimate for GHI at this site amounts to 189 W/m2, which is equivalent to 1653 kWh/m2 per year or 4.5 kWh/m2 per day.

The long-term best estimate for DNI at this site amounts to 124 W/m2, which is equivalent to 1090 kWh/m2 per year or 2.9 kWh/m2 per day.

www.suntrace.de

Page 124 of 153

5.3.2.1 GHI

For site VNDAN, six different input data sets have been assessed, from which finally the one from SG adjusted by on-site measurements was taken into account for determining the long-term annual best estimate of GHI. The main results can be found in Table 39 and Figure 107, which visualises the GHI long-term averages for these eight datasets together with the best estimate (green line).

Table 39: Overview of long-term averages from various DNI datasets.

Figure 107: Overview of the long-term GHI values from different data sets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 125 of 153

1.1.1.2 DNI

For site VNDAN, five different input datasets have been assessed, from which the SG-dataset (adjusted by on-site measurements) was taken into account for determining the long-term annual best estimate of GHI. Main results can be found in Table 40 and Figure 108, which visualises the GHI long-term averages for these five datasets together with the best estimate (green line).

Table 40: Overview of long-term averages from various DNI data sets.

Figure 108: Overview of the long-term DNI values from different datasets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 126 of 153

5.3.3 Analysis of uncertainty

As mentioned in chapter 5.1 Methodology, the total long-term uncertainty of long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainty of GHI measured by the Secondary Standard Pyranometer is assumed to be 1.6 % including the uncertainty due to calibration and maintenance. The long-term instrument uncertainty of DNI measured by the Rotating Shadowband Pyranometer is assumed to be 2.5 % including the uncertainty due to calibration and maintenance. For this study the uncertainty of the adaption methodology is assumed to be 2 % for GHI and 3 % for DNI. Based on the adapted SG data set, the inter-annual variability of GHI is 10 W/m2, approximately equivalent to ± 5.6 %. Compared to other regions in the world, this is a quite high volatility for GHI. The inter-annual variability of DNI is with 13 W/m2, approximately equivalent to ± 10.4 %, very high.

Based on the above-mentioned values, the total uncertainty of the long-term best estimates of GHI (P50) is found to be 3.0 % associated with the inter-annual variability related to multiple years and 6.3 % related to a single year. The total uncertainty of the long-term best estimates of DNI (P50) is found to be 4.9 % associated with the inter-annual variability related to multiple years and 11.5 % related to a single year. Further details can be found in Table 41.

Table 41: Explanation of derivation of the total long-term uncertainty of GHI and DNI following (Meyer, Butron, et al. 2008)

Calculation of long-term uncertainty of long-term average of GHI and DNI for multiple year and single year

overall uncertainty adjustment uncertainty

uncertainty due to temporal coverage on-site meas.

multiple year

single year

multiple year single year A B Cmy=

Csy/√n Csy Umy=√(A2+B2+Cmy

2) Usy=√(Umy

2+Csy

2) abs. [W/m2

]

rel. [%]

abs. [W/m2

]

rel. [%] [%] [%] [%] [%]

best estimate (P50) GHI

6 3.0 12 6.3 2 1.6 1.2 5.6

best estimate (P50) DNI

6 4.9 14 11.5 3 2.5 2.3 10.4

www.suntrace.de

Page 127 of 153

5.3.4 Annual cycle of GHI and DNI

Table 42 and Figure 109 for GHI and Table 43 and Figure 110 for DNI, respectively, show the long-term annual cycle of the best estimate (SGadjusted) for GHI and DNI in comparison to the original SG dataset, as well as the datasets from CMSAF-SARAH, 3Tier (GHI), DLR-ISIS, NASA-SSE and MN7.

Table 42: Long-term monthly average values of GHI.

Figure 109: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 128 of 153

Table 43: Long-term monthly average values of DNI.

Figure 110: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 129 of 153

5.4 VNCEH

5.4.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data

Figure 111 shows a scatter plot of original and adjusted SG GHI and DNI values against the on-site measurements in hourly time resolution for the overlapping period from 1 October 2017 up to 30 November 2019.

During this overlapping period, the GHI on-site measurement data had an average of 218 W/m2, while satellite-derived SG data had an average of 214 W/m2. Thus, the original SG dataset underestimates in total 1.8 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG GHI data by on-site measurements, the average of the adjusted GHI data amounts also to 218 W/m2. The DNI on-site measurement data had an average of 158 W/m2 during the overlapping period, while satellite-derived SG data had an average of 164 W/m2. Thus, the original SG data set overestimates in total 3.8 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG DNI data by on-site measurements the average of the adjusted DNI data amounts to 155 W/m2.

The frequency distribution of adjusted SG satellite-derived GHI matches very well to ground-measured values, for DNI not, as can be seen in Figure 112. However, Figure 113 shows that the monthly averages of both adjusted GHI and adjusted DNI satellite data fit better to the on-site measurements in comparison to the original satellite data for the overlapping period.

Figure 111: Scatter plots of GHI (left) and DNI (right) displaying on-site-measured hourly data points against original and adjusted GHI and DNI from SG. The black line represents the 1:1 relationship.

www.suntrace.de

Page 130 of 153

Figure 112: Frequency distribution of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 113: Monthly averages of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

5.4.2 Determination of long-term average

The long-term best estimate for GHI at this site amounts to 220 W/m2, which is equivalent to 1927 kWh/m2 per year or 5.3 kWh/m2 per day.

The long-term best estimate for DNI at this site amounts to 154 W/m2, which is equivalent to 1349 kWh/m2 per year or 3.7 kWh/m2 per day.

www.suntrace.de

Page 131 of 153

5.4.2.1 GHI

For site VNCEH, six different input data sets have been assessed, from which finally the one from SG adjusted by on-site measurements was taken into account for determining the long-term annual best estimate of GHI. The main results can be found in Table 44 and Figure 114, which visualises the GHI long-term averages for these eight datasets together with the best estimate (green line).

Table 44: Overview of long-term averages from various DNI datasets.

Figure 114: Overview of the long-term GHI values from different data sets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 132 of 153

1.1.1.3 DNI

For site VNCEH, five different input datasets have been assessed, from which the SG-dataset (adjusted by on-site measurements) was taken into account for determining the long-term annual best estimate of GHI. Main results can be found in Table 45 and Figure 115, which visualises the GHI long-term averages for these five datasets together with the best estimate (green line).

Table 45: Overview of long-term averages from various DNI data sets.

Figure 115: Overview of the long-term DNI values from different datasets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 133 of 153

5.4.3 Analysis of uncertainty

As mentioned in chapter 5.1 Methodology, the total long-term uncertainty of long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainty of GHI measured by the Secondary Standard Pyranometer is assumed to be 1.6 % including the uncertainty due to calibration and maintenance. The long-term instrument uncertainty of DNI measured by the First Class Pyrheliometer is assumed to be 2.1 % including the uncertainty due to calibration and maintenance. For this study the uncertainty of the adaption methodology is assumed to be 2 % for GHI and 3 % for DNI. Based on the adapted SG data set, the inter-annual variability of GHI is 9 W/m2, approximately equivalent to ± 3.9 %. Compared to other regions in the world, this is a normal volatility for GHI. The inter-annual variability of DNI is with 14 W/m2, approximately equivalent to ± 9.2 %, quite high.

Based on the above-mentioned values, the total uncertainty of the long-term best estimates of GHI (P50) is found to be 2.8 % associated with the inter-annual variability related to multiple years and 4.8 % related to a single year. The total uncertainty of the long-term best estimates of DNI (P50) is found to be 4.5 % associated with the inter-annual variability related to multiple years and 10.3 % related to a single year. Further details can be found in Table 46.

Table 46: Explanation of derivation of the total long-term uncertainty of GHI and DNI following (Meyer,

Butron, et al. 2008)

Calculation of long-term uncertainty of long-term average of GHI and DNI for multiple year and single year

overall uncertainty adjustment uncertainty

uncertainty due to temporal coverage on-site meas.

multiple year

single year

multiple year single year A B Cmy=

Csy/√n Csy Umy=√(A2+B2+Cmy

2) Usy=√(Umy

2+Csy

2) abs. [W/m2

]

rel. [%]

abs. [W/m2

]

rel. [%] [%] [%] [%] [%]

best estimate (P50) GHI

6 2.8 11 4.8 2 1.6 0.9 3.9

best estimate (P50) DNI

7 4.5 16 10.3 3 2.1 2.1 9.2

www.suntrace.de

Page 134 of 153

5.4.4 Annual cycle of GHI and DNI

Table 47 and Figure 116 for GHI and Table 48 and Figure 117 for DNI, respectively, show the long-term annual cycle of the best estimate (SGadjusted) for GHI and DNI in comparison to the original SG dataset, as well as the datasets from CMSAF-SARAH, 3Tier (GHI), DLR-ISIS, NASA-SSE and MN7.

Table 47: Long-term monthly average values of GHI.

Figure 116: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 135 of 153

Table 48: Long-term monthly average values of DNI.

Figure 117: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 136 of 153

5.5 VNSOB

5.5.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data

Figure 118 shows a scatter plot of original and adjusted SG GHI and DNI values against the on-site measurements in hourly time resolution for the overlapping period from 1 October 2017 up to 30 November 2019.

During this overlapping period, the GHI on-site measurement data had an average of 230 W/m2, while satellite-derived SG data had an average of 235 W/m2. Thus, the original SG dataset overestimates in total 2.2 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG GHI data by on-site measurements, the average of the adjusted GHI data amounts also to 230 W/m2. The DNI on-site measurement data had an average of 173 W/m2 during the overlapping period, while satellite-derived SG data had an average of 206 W/m2. Thus, the original SG data set overestimates in total 19.1 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG DNI data by on-site measurements the average of the adjusted DNI data amounts to 171 W/m2.

The frequency distribution of adjusted SG satellite-derived GHI matches very well to ground-measured values, for DNI not, as can be seen in Figure 119. However, Figure 120 shows that the monthly averages of both adjusted GHI and adjusted DNI satellite data fit better to the on-site measurements in comparison to the original satellite data for the overlapping period.

Figure 118: Scatter plots of GHI (left) and DNI (right) displaying on-site-measured hourly data points against original and adjusted GHI and DNI from SG. The black line represents the 1:1 relationship.

www.suntrace.de

Page 137 of 153

Figure 119: Frequency distribution of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 120: Monthly averages of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

5.5.2 Determination of long-term average

The long-term best estimate for GHI at this site amounts to 226 W/m2, which is equivalent to 1977 kWh/m2 per year or 5.4 kWh/m2 per day.

The long-term best estimate for DNI at this site amounts to 158 W/m2, which is equivalent to 1385 kWh/m2 per year or 3.8 kWh/m2 per day.

www.suntrace.de

Page 138 of 153

5.5.2.1 GHI

For site VNSOB, six different input data sets have been assessed, from which finally the one from SG adjusted by on-site measurements was taken into account for determining the long-term annual best estimate of GHI. The main results can be found in Table 49 and Figure 121, which visualises the GHI long-term averages for these eight datasets together with the best estimate (green line).

Table 49: Overview of long-term averages from various DNI datasets.

Figure 121: Overview of the long-term GHI values from different data sets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 139 of 153

1.1.1.4 DNI

For site VNSOB, five different input datasets have been assessed, from which the SG-dataset (adjusted by on-site measurements) was taken into account for determining the long-term annual best estimate of GHI. Main results can be found in Table 50 and Figure 122, which visualises the GHI long-term averages for these five datasets together with the best estimate (green line).

Table 50: Overview of long-term averages from various DNI data sets.

Figure 122: Overview of the long-term DNI values from different datasets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 140 of 153

5.5.3 Analysis of uncertainty

As mentioned in chapter 5.1 Methodology, the total long-term uncertainty of long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainty of GHI measured by the Secondary Standard Pyranometer is assumed to be 1.6 % including the uncertainty due to calibration and maintenance. The long-term instrument uncertainty of DNI measured by the Rotating Shadowband Pyranometer is assumed to be 2.5 % including the uncertainty due to calibration and maintenance. For this study the uncertainty of the adaption methodology is assumed to be 2 % for GHI and 3 % for DNI. Based on the adapted SG data set, the inter-annual variability of GHI is 9 W/m2, approximately equivalent to ± 4.0 %. Compared to other regions in the world, this is a normal volatility for GHI. The inter-annual variability of DNI is with 18 W/m2, approximately equivalent to ± 11.4 %, quite high.

Based on the above-mentioned values, the total uncertainty of the long-term best estimates of GHI (P50) is found to be 2.8 % associated with the inter-annual variability related to multiple years and 4.9 % related to a single year. The total uncertainty of the long-term best estimates of DNI (P50) is found to be 5.0 % associated with the inter-annual variability related to multiple years and 12.4 % related to a single year. Further details can be found in Table 51.

Table 51: Explanation of derivation of the total long-term uncertainty of GHI and DNI following (Meyer, Butron, et al. 2008)

Calculation of long-term uncertainty of long-term average of GHI and DNI for multiple year and single year

overall uncertainty adjustment uncertainty

uncertainty due to temporal coverage on-site meas.

multiple year

single year

multiple year single year A B Cmy=

Csy/√n Csy Umy=√(A2+B2+Cmy

2) Usy=√(Umy

2+Cs

y2)

abs. [W/m2]

rel. [%]

abs. [W/m2]

rel. [%] [%] [%] [%] [%]

best estimate (P50) GHI

6 2.8 11 4.9 2 1.6 0.9 4.0

best estimate (P50) DNI

8 5.0 20 12.4 3 2.5 2.5 11.4

www.suntrace.de

Page 141 of 153

5.5.4 Annual cycle of GHI and DNI

Table 52 and Figure 123 for GHI and Table 53 and Figure 124 for DNI, respectively, show the long-term annual cycle of the best estimate (SGadjusted) for GHI and DNI in comparison to the original SG dataset, as well as the datasets from CMSAF-SARAH, 3Tier (GHI), DLR-ISIS, NASA-SSE and MN7.

Table 52: Long-term monthly average values of GHI.

Figure 123: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 142 of 153

Table 53: Long-term monthly average values of DNI.

Figure 124: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 143 of 153

5.6 VNTRA

5.6.1 Comparison and adjustment of satellite derived GHI and DNI values to on-site measurement data

Figure 125 shows a scatter plot of original and adjusted SG GHI and DNI values against the on-site measurements in hourly time resolution for the overlapping period from 1 October 2017 up to 30 November 2019.

During this overlapping period, the GHI on-site measurement data had an average of 220 W/m2, while satellite-derived SG data had an average of 216 W/m2. Thus, the original SG dataset underestimates in total 1.8 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG GHI data by on-site measurements, the average of the adjusted GHI data amounts also to 220 W/m2. The DNI on-site measurement data had an average of 158 W/m2 during the overlapping period, while satellite-derived SG data had an average of 159 W/m2. Thus, the original SG data set underestimates in total 0.6 % in comparison to the on-site measurements for the overlapping period. After an adjustment of satellite-derived SG DNI data by on-site measurements the average of the adjusted DNI data amounts also to 158 W/m2.

The frequency distribution of adjusted SG satellite-derived GHI and DNI matches very well to ground-measured values as can be seen in Figure 126. Figure 127 shows the Monthly averages of GHI and DNI for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 125: Scatter plots of GHI (left) and DNI (right) displaying on-site-measured hourly data points against original and adjusted GHI and DNI from SG. The black line represents the 1:1 relationship.

www.suntrace.de

Page 144 of 153

Figure 126: Frequency distribution of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

Figure 127: Monthly averages of GHI (left) and DNI (right) for original and adjusted satellite data from SG, as well as for on-site-measured values for the overlapping period.

5.6.2 Determination of long-term average

The long-term best estimate for GHI at this site amounts to 218 W/m2, which is equivalent to 1909 kWh/m2 per year or 5.2 kWh/m2 per day.

The long-term best estimate for DNI at this site amounts to 152 W/m2, which is equivalent to 1334 kWh/m2 per year or 3.7 kWh/m2 per day.

www.suntrace.de

Page 145 of 153

5.6.2.1 GHI

For site VNTRA, six different input data sets have been assessed, from which finally the one from SG adjusted by on-site measurements was taken into account for determining the long-term annual best estimate of GHI. The main results can be found in Table 54 and Figure 128, which visualises the GHI long-term averages for these eight datasets together with the best estimate (green line).

Table 54: Overview of long-term averages from various DNI datasets.

Figure 128: Overview of the long-term GHI values from different data sets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 146 of 153

1.1.1.5 DNI

For site VNTRA, five different input datasets have been assessed, from which the SG-dataset (adjusted by on-site measurements) was taken into account for determining the long-term annual best estimate of GHI. Main results can be found in Table 55 and Figure 129, which visualises the GHI long-term averages for these five datasets together with the best estimate (green line).

Table 55: Overview of long-term averages from various DNI data sets.

Figure 129: Overview of the long-term DNI values from different datasets (red: used / blue: not used for determination of best estimate) and the best estimate (green line). The lower (cyan) and upper (magenta) lines illustrate the ranges of the best estimate, which are based on the estimated uncertainty of the best estimate.

www.suntrace.de

Page 147 of 153

5.6.3 Analysis of uncertainty

As mentioned in chapter 5.1, Methodology the total long-term uncertainty of long-term best estimate of GHI and DNI depends on the uncertainty of measurements, uncertainty of long-term data and their inter-annual variability. The long-term instrument uncertainty of GHI measured by the Secondary Standard Pyranometer is assumed to be 1.6 % including the uncertainty due to calibration and maintenance. The long-term instrument uncertainty of DNI measured by the First Class Pyrheliometer is assumed to be 2.1 % including the uncertainty due to calibration and maintenance. For this study the uncertainty of the adaption methodology is assumed to be 2 % for GHI and 3 % for DNI. Based on the adapted SG data set, the inter-annual variability of GHI is 8 W/m2, approximately equivalent to ± 3.8 %. Compared to other regions in the world, this is a normal volatility for GHI. The inter-annual variability of DNI is with 13 W/m2, approximately equivalent to ± 8.5 %, quite high.

Based on the above-mentioned values, the total uncertainty of the long-term best estimates of GHI (P50) is found to be 2.8 % associated with the inter-annual variability related to multiple years and 4.7 % related to a single year. The total uncertainty of the long-term best estimates of DNI (P50) is found to be 4.4 % associated with the inter-annual variability related to multiple years and 9.6 % related to a single year. Further details can be found in Table 56.

Table 56: Explanation of derivation of the total long-term uncertainty of GHI and DNI following (Meyer,

Butron, et al. 2008)

Calculation of long-term uncertainty of long-term average of GHI and DNI for multiple year and single year

overall uncertainty adjustment uncertainty

uncertainty due to temporal coverage on-site meas.

multiple year

single year

multiple year single year A B Cmy=

Csy/√n Csy Umy=√(A2+B2+Cmy

2) Usy=√(Umy

2+Cs

y2)

abs. [W/m2]

rel. [%]

abs. [W/m2]

rel. [%] [%] [%] [%] [%]

best estimate (P50) GHI

6 2.8 10 4.7 2 1.6 0.9 3.8

best estimate (P50) DNI

7 4.4 15 9.6 3 2.1 1.9 8.5

www.suntrace.de

Page 148 of 153

5.6.4 Annual cycle of GHI and DNI

Table 57 and Figure 130 for GHI and Table 58 and Figure 131 for DNI, respectively, show the long-term annual cycle of the best estimate (SGadjusted) for GHI and DNI in comparison to the original SG dataset, as well as the datasets from CMSAF-SARAH, 3Tier (GHI), DLR-ISIS, NASA-SSE and MN7.

Table 57: Long-term monthly average values of GHI.

Figure 130: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 149 of 153

Table 58: Long-term monthly average values of DNI.

Figure 131: Long-term monthly averages of GHI best estimate (SGadjusted) in comparison to SGoriginal, CMSAF, DLR-ISIS, NASA-SSE and MN7 datasets.

www.suntrace.de

Page 150 of 153

6 Conclusions

In this report, we analysed 24+ months of measurement data recorded at five different sites; VNHAN, VNDAN, VNCEH, VNSOB and VNTRA. The station network comprises solar measurement stations of type Tier 1 and Tier 2. The data presented in this document was recorded from Septermber 2017 to November 2019. Today, stations are still operating. During this period, the international expert Joana Zerbin from Suntrace conducted 2 visits to all five stations. During these visits, the expert tested each station for functionality and performed a visual inspection. For all stations, station keepers participated in refreshment trainings, during which instructions were given on station maintenance and documentation. During year 1, an outage of the RSI at VNSOB occurred for which Suntrace compensated by an additional 2 months of measurement (October and November 2019). At station VNCEH, broken cables and false cabling caused data loss for soiling analysis until March 2018. During year 2, malfunction of the network concerned station maintenance by the station keeper at VNSOB: Due to false cleaning, Reference Cell 2 could not be utilised for soiling analysis. However, Refereence Cell 3 held necessary information nonetheless. Furthermore, insufficient documentation at station VNSOB must be reported due to a lack of supplied sheets.

Table 59: Overview of solar and auxiliary measurements

Parameter UNIT VNHAN VNDAN VNCEH VNSOB VNTRA

GHI kWh/m2/day W/m2

3.6 150

4.5 188

5.2 218

5.5 229

5.3 221

DNI kWh/m2/day W/m2

1.8 73

3.1 128

3.8 158

4.3 181

3.8 158

DHI kWh/m2/day W/m2

2.3 97

2.3 95

2.5 102

2.2 91

2.4 101

Temperature °C 31.7 32.6 34.6 30.8 34.9

humidity % 79 85 74 73 81

Precipitation * mm – – – 366 –

Air pressure hPa 1005 1004 979 1002 1001

wind gusts m/s 25.7 20.3 26.0 20.4 25.1

Only full meteorological years considered: Average values from 1 October 2017 to 30 September 2019 *Annual cummulative precipitation

www.suntrace.de

Page 151 of 153

Despite the stated deficiencies, the measurement period held valuable information for the assessment of Vietnam’s solar resource. Table 59 lists averages of solar and auxiliary parameters measured at the five stations. The table reveals a clear increase in irradiance towards the South. Thus, we can find highest yields at stations VNCEH, VNSOB and VNTRA. This trend is illustrated in Figure 132.

Figure 132: Long-term annual average in DNI and GHI dervied from satellite-based data. Based satellite-derived GHI and DNI, the long-term best estimes (P50) were calculated for each station and compared to the ground-based biannual recordings. Thus, each location data availability encompasses 20+ years of satellite-derived data and 2+ years of on-site data. Table 60 summarises the long-term estimates for GHI and DHI that are to be found at each station. Long-term and on-site measurements correlate well, as can be concluded from comparing Table 59 and Table 60. Both reveal daily yields to lie above 5 kWh/m2 at stations VNSOB, VNCEH and VNTRA.

Table 60: Overview of long-term GHI and DNI estimates

Parameter UNIT VNHAN VNDAN VNCEH VNSOB VNTRA

GHI kWh/m2/day W/m2

3.6 148

4.5 189

5.3 220

5.4 226

5.2 218

DNI kWh/m2/day W/m2

1.8 74

3.0 124

3.9 161

3.8 158

3.6 152

0

50

100

150

200

250

VNHAN VNDAN VNCEH VNSOB VNTRA

long

-ter

m a

nnua

l ave

rage

[W/m

2]

GHI DNI

www.suntrace.de

Page 152 of 153

7 Summary

• During the monitored 24 months, highest mean values recorded at southernmost stations VNSOB and VNTRA

• Very low direct irradiance at VNHAN (nearby Hanoi) • Long-term GHI and DNI predictions correlate with in-situ measurements • Hot and humid climate: average relative humidity often exceeds 80 %, annual

temperature above 30 °C • Precipitation: high intra-annual variability with extreme rainfall events (monsoon) • Impact by wind: 5 % of wind data exceeded 7.7 m/s; wind gusts may reach 26.0

m/s • Soiling loss depends on season (self-cleaning during rainy season) and may

exceed 5 % during dry season • Only short timespans available to estimate soiling losses

www.suntrace.de

Page 153 of 153

8 Bibliography

Meyer, R, H G Beyer, J Fanslau, N Geuder, A Hammer, T Hirsch, C Hoyer-Click, N Schmidt, and M Schwandt. 2009. Towards standardization of CSP yield assessments. http://solarpaces2009.org.

Meyer, R, J T Butron, G Marquardt, M Schwandt, N Geuder, C Hoyer-Klick, E Lorenz, and Hammer A. 2008. Combining solar irradiance measurements and various satellite-derived products to a site specific best estimate.

Mieslinger, T, F Ament, K Chhatbar, and M Meyer. 2014. “A new method for fusion of measured and model-derived solar radiation time-series.” Science Direct 1617 - 1626.


Recommended