Time Series Model of Photovoltaic Generation for Distribution Planning Analysis Jorge Valenzuela
Overview Introduction: The solar problem and our limitations Modeling What information do we have? Solar Irradiation characteristics Simple Model & Time Series Error analysis Examples of use Light load problems Peak conditions Conclusions and Comments Outlook 2
Introduction - The solar problem Current deployment rate in Massachusetts: From 2007 solar capacity increased from 7MW to 250 MW Goal: 1,600 MW of PV installed in Massachusetts by 2020 In 2012, 34 MW of residential PV was added (NG territory) 300 Histogram of PV Installations (NG) 120% Residential PV capacity installed by year (NG) 250 100% 40 35 30 200 80% [MW] 25 20 15 10 Frequency 150 100 60% 40% 5 0 50 20% 2007 2008 2009 2010 2011 2012 year 0 0% 0 2 3 4 5 6 7 8 9 10 12 16 25 53 120 270 750 Bin (Capacity kw) 3
Introduction - Our limitations Visibility: Real time generation Geographical location What level of generation should I use during my long term studies? How will PV installations affect Peak and Light load conditions? What kind of voltage effects will the PV units have on my feeders? What kind of impact the future levels of PV penetration will have on my feeders? What effect would PV units have on DG limits during switching events?? Decisions need to be made for the next 5, 10, 15 years! 4
Solar generation - What we have Sutton/Northbridge Solar Power Project 4,683 solar panels, 983 kw (DC) Real Time generation monitoring Irradiation and Temperature monitoring Power Generated [kw] 900 800 700 600 500 400 300 200 100 0 May June July Aug Sept Oct Nov Dec Jan Feb Mar Apr May Month National Grid Delivery Services, Sutton/Northbridge Solar Power Project, (http://www.nationalgridus.com/masselectric/solar/sutton.asp, 2013, [Accessed June 2013]) 5
Solar irradiation - Characteristics I I * cos( q) I I T D d r I A e ; I C *I * F ; I I * * F -B/sin( b ) D d D SS r th g sg I D : Direct irradiation [W/m 2 ] I d : Diffuse irradiation [W/m 2 ] I r : Reflected irradiation [W/m 2 ] q: Incident angle A: Apparent solar irradiation coefficient B: Atmospheric extinction coefficient b: Solar altitude angle C: Ratio of diffuse radiation on a horizontal surface to the direct solar radiation F SS : Angle factor between surface and sky F sg : Angle factor between surface and ground I th : Total horizontal irradiation American Society of Heating, Refrigerating and Air-Conditioning Engineers, 2011 ASHRAE Handbook - Heating, Ventilating, and Air-Conditioning Applications (I-P Edition), (ISBN: 978-1-936504-06-0, Apr 2012) 6
Solar irradiation - Modeling I I * cos( q) I I ; I A e ; I C *I *F ; I I * *F -B/sin( b) T D d r D d D SS r th g sg b: Solar altitude angle q: Incident angle cos(θ)=sin(β)*cos(t)+cos(β)*sin(t)*cos(α 1-α 2) sin(β)=cos(lat)*cos(δ)*cos(h)+sin(lat)*sin(δ) α =cos 1-1 (sin(β)*sin(lat)-sin(δ)) *sgn(h) cos(β)*cos(lat) a1: Solar azimuth angle o H=((Time(mins)-720)/4)*15 o o δ=23.45 *sin 360 * 284+N /365 a2: Azimuth angle of solar panel 0 if facing South) LAT: Latitude angle T: Tilt of the solar panel (set to LAT) H: Number of Sun hours d: Declination angle N: Day Number (1=January 1st) 7
Solar irradiation - Modeling -1 SS=720min ± cos (-tan(l)*tan(δ))*4min/deg 365 N=1-1 -1 SV(t)= u t- 1440*N-720-cos (-tan(l)*tan(δ))*4 + u t- 1440*N-720+cos (-tan(l)*tan(δ))*4 UO SRML: Sun path chart program [online] (Available from: http://solardat.uoregon.edu/sunchartprogram.html, 2007, [Accessed June 2013]) 8
Solar irradiation - Model comparison I A e * cos( q) C * F + I * * F * -B/sin( b) T SS th g sg 365 N=1-1 -1 u t- 1440*N-720-cos (-tan(l)*tan(δ))*4 + u t- 1440*N-720+cos (-tan(l)*tan(δ))*4 A: Apparent solar irradiation coefficient B: Atmospheric extinction coefficient C: Ratio of diffuse radiation on a horizontal a horizontal surface to the direct solar radiation q: Incident angle b: Solar altitude angle F SS : Angle factor between surface and sky F sg : Angle factor between surface and ground I th : Total horizontal irradiation 1 Irradiation/Generation (p.u.) 0.9 0.8 0.7 0.6 0.5 0.4 0.3 Calculated Measured 0.2 May Jun Jul Aug Sep Oct Nov Dec Jan Feb Mar Month 9
Solar irradiation - Error decomposition R. B. Cleveland, W. S. Cleveland, J.E. McRae, and I. Terpenning, STL: A Seasonal-Trend Decomposition Procedure Based on Loess, (Journal of Official Statistics, 6, 3 73, 1990) 10
Solar generation - Simplified model P =P * 0.9*I -0.23*sin(2*π*(t+14400)/1051200)*rand(t) *SV(t) g Nom T 365 N=1-1 -1 SV(t)= u t- 1440*N-720-cos (-tan(l)*tan(δ))*4 + u t- 1440*N-720+cos (-tan(l)*tan(δ))*4 Solar Power Generation (p.u.) 1 Model (p.u.) 0.9 Measured (p.u) 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0 Jul-17 Jul-18 Jul-19 Jul-20 Jul-21 Jul-22 Jul-23 0.5 Measured (p.u.) Model (p.u.) Solar Power Generation (p.u.) 0.4 0.3 0.2 0.1 0 Dec-17 Dec-18 Dec-19 Dec-20 Dec-21 Dec-22 Dec-23 11
Examples of use Solar Power Generation and Feeder Loading (p.u.) 1.1 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 Summer Peak Solar Generation ~ 45% at peak Light Load May 9 Solar Generation ~ 90% near noon Measured (p.u.) Feeder Load (p.u.) 1 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 Feeder Load (p.u.) 01-May 00:00 27-May 00:59 22-Jun 01:59 18-Jul 02:59 13-Aug 03:59 08-Sep 04:58 04-Oct 05:58 30-Oct 06:58 25-Nov 07:58 21-Dec 08:58 16-Jan 09:57 11-Feb 10:57 08-Mar 11:57 In New England, the effects on feeders from solar generation during May can be bigger than during Summer time Solar Power Generation and Feeder Loading (p.u.) 1 0.8 0.6 0.4 0.2 0 Light Load May 9 Solar Generation ~ 90% Measured (p.u.) Model (p.u.) Feeder Load (p.u.) 08-May 06:14 08-May 18:14 09-May 06:14 09-May 18:14 10-May 06:14 10-May 18:14 11-May 06:14 11-May 18:14 12-May 06:14 12-May 18:14 13-May 06:14 0.65 0.6 0.55 0.5 0.45 0.4 Feeder Load (p.u.) 12
Examples of use Also, during annual coincidental peak time, solar generation gets close to 45% of the nameplate Solar Power Generation (p.u.) 1.2 1 0.8 0.6 0.4 0.2 Model (p.u.) Measured (p.u) Feeder Load (p.u.) 0 Jul-17 Jul-18 Jul-19 Jul-20 Jul-21 Jul-22 Jul-23 Peak For this particular area, capacity planning analysis (Peak) can be done considering PV generation of up to 45% (even a 65% can be used as a conservative number) Depending on the study, different loading levels can be chosen. A clear sky value would give the most reasonable value for the maximum expected solar generation 13
Conclusions and Comments A procedure to create a solar generation model was presented. The model can bring more clarity to planners by estimating generation of solar units that are not visible on the system The model, based on solar irradiation, can be easily customized to the location of analysis Local data of solar generation, if available, can be used to calibrate the model. This requires error analysis Once implemented for a certain region, it can be saved as a lookup table for easy reference Clear sky values of generation can be used for planners to do their analysis during peak and light load conditions. The same can be applied to DG limits on feeders during temporary switching Analysis of the error between theoretical values and local measurements presents an opportunity for modeling (stochastic) of cloud coverage 14
Outlook More data! 15
Thank you! 16