Students' Senior Project: Ali Salman, Hussain A.Ali, Alaa Jameel Supervised by:

Dr. Mohamed Bin Shams Dr. Shaker Haji

The Sixth Zayed Seminar Arabian Gulf University May 8-9, 2013 – Kingdom of Bahrain

2

Feb 21, 2010

Visitors Center Hall: 10 x 60 W light bulbs + a 2 ton A/C Total ~ 3 kW

2 Renewable Energy Sources

4 x 260Ah/12V (48 V DC System - Series)

20 x 200 Wp = 4 kWp

1.7 kW @ 12 m/s cut in: 3 m/s

1.2 kW Nexa FC Stack Public Grid

6 x 500 NL @ 10 bar H2 M-H Canisters

2 x 60 NL/h H2 Generator (water electrolyzers)

Bapco’s Green Energy Station Configuration. 5

Simplified Power Management Strategy [1]

6

Climate Related Data Environmental Conditions

50

900

45

800

40

700

35

600

30

500

Solar Irradiation

25

400

Ambient Temperature

20

300

Module Temperature

15

200

Wind Velocity

10

100

5

0

0

12:00 AM 24/06/2010

6:00 AM

12:00 PM

6:00 PM

Date & Time

12:00 AM 25/06/2010

Temperature [oC] or Wind Velocity [m/s]

Power [W/m2]

1000

6:00 AM 7

Courtesy of Bapco

Energy Related Data

3000

100

AC Laod P [W] 2500

WT P [W]

80

FC P [W]

Power [W]

2000

70

Bat. SOC [%] 60

H2 Pressure 1500

50 40

1000

30 20

500

Battery SOC [%] or H2 Pressure [bar]

90

PV P [W]

10 0 12:00 AM 24/06/2010

0 6:00 AM

12:00 PM

6:00 PM

Date & Time

12:00 AM

6:00 AM

25/06/2010 8

Courtesy of Bapco

9

What does the performance of a Hybrid Renewable Energy System depend on? Energy Generators

Power Demand

Climate Conditions

Energy Storage Performance

10

(1) To analyze, model, & forecast meteorological data (driving forces) (2) To analyze the energy data from the hybrid RE system Ultimate Goal: Optimize the performance of the RE system (e.g. maximizing CO2 saving)

11

Data collection (1st May 2010 – 30th April 2011)

Bapco’s Green Energy Station Data.

Objective 1: Time Series Analysis:

• Why do we use Time Series Analysis? - Data are correlated in time

Time Series Analysis: 𝒚𝒚𝒕𝒕 = 𝜹𝜹 + Ф𝟏𝟏 𝒚𝒚𝒕𝒕−𝟏𝟏 + Ф𝟐𝟐 𝒚𝒚𝒕𝒕−𝟐𝟐 + ⋯ + Ф𝒑𝒑 𝒚𝒚𝒕𝒕−𝒑𝒑 + 𝜺𝜺𝒕𝒕 − 𝜽𝜽𝟏𝟏 𝜺𝜺𝒕𝒕−𝟏𝟏 − 𝜽𝜽𝟐𝟐 𝜺𝜺𝒕𝒕−𝟐𝟐 − ⋯ − 𝜽𝜽𝒒𝒒 𝜺𝜺𝒕𝒕−𝒒𝒒

AR: Auto Regressive

MA: Moving Average ARMA

15

Time Series Analysis: . Data should be stationary.

Stationary

Const. Mean

Const. Variance

Time Series Analysis :

• Transformations: ARIMA (p,I,q) – (const. mean) – (const. variance)

Differencing Box – Cox Transformation

Mean Stabilization

Differencing Transformation [3]

Variance Stabilization

BOX-COX Transformation [3]

Time Series Analysis 1. Estimating Missing Points 2. Stationarizing data (if required) 3. Parameter Estimation (Least Square Error) 4. Validating Models (Residual Analysis) 5. Forecasting 6. Validating Forecasting Accuracy 7. Other Statistical Analysis

Energy Data Analysis 1. Efficiency Calculations: various systems 2. Energy Calculations: Load, RE, Storage & PG 3. Calculations of CO2 Savings & Emissions 4. Feasibility Calculations: PV & WT

Speed, m/s

• Average daily wind speed: ARIMA (1,0,0): 53 42 42 31 31 20 -1 2 -1 1 -2 1 -2 0

Actual Data Actual wind speed Forcasted Trans. Actualwind wind Datavelocity speed ModeledUCL wind speed

𝒚𝒚 =

(𝒙𝒙)𝟎𝟎.𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑−𝟏𝟏 𝟎𝟎.𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑𝟑

𝒚𝒚𝒕𝒕 = 𝟎𝟎. 𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏𝟏 + 𝟎𝟎. 𝟔𝟔𝟔𝟔𝟔𝟔𝟔𝟔𝟔𝟔 𝒚𝒚𝒕𝒕−𝟏𝟏 + 𝜺𝜺𝒕𝒕

LCL

• Average daily solar radiation: ARIMA (1,0,0): Radiation W/m2 Radaition W/m2

600 500 400 300 200 100 0

Actual Solar Radiation UCL

LCL

Modeled Solar Radiation Forecasted Data

Actual Data

𝒚𝒚𝒕𝒕 = 𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖𝟖 + 𝟎𝟎. 𝟓𝟓𝟓𝟓𝟓𝟓𝟓𝟓𝟓𝟓 𝒚𝒚𝒕𝒕−𝟏𝟏 + 𝜺𝜺𝒕𝒕

• Average daily ambient temperature: ARIMA (0,1,2): 45 40.00 Temperature, Temperature, ̊C̊C

40 35.00 35 30 30.00 25 25.00 20 15 20.00 10 15.00 5

Actual Ambient Modeled Ambient Temperature Actual DataTemperature Forecasted Data UCL LCL

𝒚𝒚𝒕𝒕 = −𝟎𝟎. 𝟎𝟎𝟎𝟎𝟎𝟎 + 𝒚𝒚𝒕𝒕−𝟏𝟏 + 𝟎𝟎. 𝟐𝟐𝟐𝟐𝟐𝟐𝜺𝜺𝒕𝒕−𝟏𝟏 + 𝟎𝟎. 𝟐𝟐𝟐𝟐𝟐𝟐𝜺𝜺𝒕𝒕−𝟐𝟐 + 𝜺𝜺𝒕𝒕

Temperature, Temperature, °C°C

• Average daily solar module temperature: ARIMA (0,1,1): 50 45 45 40 40 35 30 35 25 30 20 25 15 10 20

Actual ActualData Module Temperature Forcasted Data

Modeled Module UCL TemperatureLCL

𝒚𝒚𝒕𝒕 = −𝟎𝟎. 𝟎𝟎𝟎𝟎𝟎𝟎𝟎𝟎 + 𝒚𝒚𝒕𝒕−𝟏𝟏 + 𝟎𝟎. 𝟒𝟒𝟒𝟒𝟒𝟒𝟒𝟒𝟒𝟒 𝜺𝜺𝒕𝒕−𝟏𝟏 + 𝜺𝜺𝒕𝒕

Monthly load of the station 3,500

Energy, MJ

3,000 2,500 2,000 1,500 1,000 500 0

Load of the station, MJ

Monthly systems contributions to load demand Monthly Contribution, %

100% 80% 60% 40% 20% 0%

Annual and systems contributions to load demand 0 0%

Monthly Contribution of Renewable Sources, % 5.14 GJ 30%

10.7 GJ Grid Monthly Contribution of public Grid, % 90% Renewable Energy

11.95 GJ 70%

Solar system 1.25 GJ 10%

Wind Energy

Energy, MJ

Monthly energy out from storage system 1000 900 800 700 600 500 400 300 200 100 0

Annual energy out from storage system 236 MJ 5% Storage Energy Out from Batteries, MJ 4268 MJ 95%

Annual Storage Energy Out from Batteries Storage Energy Out from Fuel Cell, MJ Annual Storage Energy Out from Fuel Cell

• CO2 Savings: 250

CO2 , kg

200 150 100 50 0

Annual Saved CO2 , kg 1,429 Estimated CO2 Savings, kg

Annual Emitted CO2 , kg CO2 Emitted, kg 621

• Units efficiencies: Equipment

Efficiency, %

Maximum

Minimum

Standard deviation

Reference

Wind Turbine

57.64% (a)

74.34%

37.29%

9.74%

< 59% (Bitz) [4]

Solar Panel

7.00% (a)

11.56%

3.15%

2.46%

< 15% max [5]

Fuel Cell

37.7% (b)

37.75%

37.65%

-

< 40-60% [6]

(a) Average monthly efficiency (b) Average instantaneous efficiency for October and February

• Effectiveness, Payback period, energy cost of Wind & Solar systems: Equip.

Cost, $/W

Lifetime, years

Rated Power, kW

Avg. Meas. Power, kW

Effectiveness, %

PBP, years Comm.

Resid.

Energy Cost, $/kWh (fils/kWh)

Wind Turbine

2.4

20

1.7

0.0482

2.83

276

1349

1.18 (442)

Solar Panel

4.2

20

4

1.1161

25.76

133.5

680

0.59 (223)

* Effectiveness = Avg. Measured Power / Rated Power

The following models were found satisfactory in modeling:

– • • • •

ARIMA(1,0,0): Wind speed ARIMA(1,0,0): Solar radiation ARIMA(0,1,2): Ambient temperature ARIMA(0,1,1): Solar module’s temperature data

–

The station load was met by renewable energy (70%) and public grid (30%).

–

Solar Panels contributed with 90% while the wind turbine contributed with 10% to the RE mix.

–

Due to the governmental subsidies of electricity, non of the PV or WT was found feasible .

–

The solar panel is more feasible than the wind turbine.

–

Modify the mechanism, so the grid cover only the energy shortage due insufficient supply from renewable and storage systems.

–

Regular maintenance and upgrading the data acquisition system.

–

Subsidies should be provided to RE technologies for them to be feasible .

–

Optimizing of the station operation/configuration, for further CO2 savings.

– Bapco, Awali Services – Prof Waheeb Al-Naser & Mr Hussain AlAnsari – Heliocentris (system manufacturer) – Dr Hanan Al-Buflasa – Our students: Ali Salman, Hussain A.Ali, Alaa Jameel.

[1] Dr.Cluas Fischer and Dr. Nroman Siehl, 2010, Heliocentris Energirsysteme GmbH, Power Point presentation, Heliocentis, Berlin-Germany [2] Montgomery etal, (2008). Introduction to Time Series Analysis and Forecasting, Wiley. River Street Hoboken [3] Description of Transformation, http://www.xlstat.com/en/learning-center/tutorials/usingdifferencing-to-obtain-a-stationary-time-series.html, [Last visit on 13 January 2013] [4] The Royal Academy of Engineering (Wind Turbine Power calculations) http://www.raeng.org.uk/education/diploma/maths/pdf/exemplars_advanced/23_wind_turbine. pdf, [Last visit on 6 January 2013] [5] Alfasolar, (Solar Module Series alfasolar pyramid 54) , http://www.alfasolar.de, [Last visit on 6 January 2013] [6] EERE Information Center-U.S (Fuel Cell Technologies Program) http://www1.eere.energy.gov/hydrogenandfuelcells/pdfs/fct_h2_fuelcell_factsheet.p df, [Last visit on 6 January 2013]