2014 PV Performance Modeling Workshop: Optimization strategies with Pvsyst for large scale PV installations: Bruno Wittmer, Pvsyst
6 likes5,768 views
The document outlines optimization strategies for large-scale photovoltaic (PV) installations using PVSyst software, emphasizing the importance of batch simulations for comprehensive analysis of design variables. It discusses various parameters, such as ground covering ratio and pitch, and their impact on economic performance, including profitability, cost of energy, and design considerations. The document concludes with a summary of findings and future outlook for enhancements in batch simulation capabilities.
2014 PV Performance Modeling Workshop: Optimization strategies with Pvsyst for large scale PV installations: Bruno Wittmer, Pvsyst
- 1. PVSYST SA - Route du Bois-de-Bay 107 - 1242 Satigny - Suisse www.pvsyst.com Any reproduction or copy of the course support, even partial, is forbidden without a written authorization of the author. Optimization strategies with Pvsyst for large scale PV installations Bruno Wittmer bruno.wittmer@pvsyst.com
- 2. Page 2Page 2 • Introduction • Batch simulations • Optimization – Basic results – Economical evaluations • Summary and Outlook
- 3. Page 3Page 3 Motivation • Optimization process is often long and tedious − Multivariate optimization − Variables can have non-intuitive effects − Often variables have complex correlations • Optimization can be driven by different figures of merit − ‘Technical’ Measures (EGrid, PR, etc. ) − Economic Measures (Returns, Payback, LCOE, etc.) • Some design variables of a PV installation can be varied continuously (‘Batch Simulations’) − This allows a more comprehensive analysis − Move from single simulation variants to batch simulations
- 4. Page 4Page 4 Reference Project • Be as specific as possible without compromising variation of batch parameters Reference Project Layout 40 sheds, 3 rows per shed Modules Generic 250 W module Inverters Generic 500 kW inverter Power 11520 modules, Pnom = 2.88 GWp Shadings According to strings ( & linear) Meteo Input Meteonorm 6.1 for Geneva No additional shading objects ! Large system
- 5. Page 5Page 5 Batch simulations • PVsyst needs a CVS file with the parameters for the simulations • Parameter filling and analysis were performed with a framework written in the R language Reference Project Parameter and Results selection Template CSV File Batch Execution Results CSV File Parameter Filling Analysis and Plotting
- 6. Page 6Page 6 Batch parameters • Several simulation parameters can be varied in the batch simulations • For this presentation only Tilt and Pitch were used • More parameters will be added in the coming versions Site and Meteo • Site • Meteo File Orientation • Tilt • Azimuth 3D Shading • Pitch N-S • Shed width System • PV module • Rserie • Rshunt • Rshunt(0) • Nr. Mod. Series • Nr. strings • Module Qlty loss • Inverter model • Nr. Inverters or MPPT
- 7. Page 7Page 7 Ground Covering Ratio (GCR) and Pitch • PVsyst will vary the pitch in the batch simulations • The plots in this presentation use the GCR • For homogeneous sheds the GCR is defined as Width/Pitch • Assuming that the system scales with the size, one can renormalize to a given area Reference Project Width 3.04 m Pitch 6.8 m GCR 45% Batch Simulation GCR 10% – 100% in steps of 2% Pitch 30.4 m – 3.04 m, variable steps
- 8. Page 8Page 8 Input and Output Variables • Input Variables added to the CSV template file: 2300 Simulations take around 3h computing time Param. Range Step Nr. steps Tilt 1° - 50° 1° 50 GCR 10% – 100% 2% 45 Pitch 30.4 m – 3.04 m variable 45 • Output as CSV file(s): − All PVsyst simulation variables can be chosen for output Between 60 and 90 variables depending on simulation type − Output is saved as yearly sums − Optionally: create hourly values for each simulation (not used here) • Output variables in this presentation: − Mostly EGrid
- 9. Page 9Page 9 What are the best GCR and Tilt? • Most simple measure is Egrid • One could also use EArray and optimize the inverter in a second step • Optimal Tilt lies on the grey line • Performance Ratio is not a good measure • Fails to recognize different incident Energy as function of Tilt • Inherent to definition of PR Optimal Tilt for given GCR
- 10. Page 10Page 10 Fixed Pnom or fixed area? • EGrid: scenario with fixed Pnom • EGrid/pitch: scenario with fixed area • Optimal Tilt line is the same for both fixed Pnom fixed area Note the different scale ‼ • GCR = 0 is not possible The surface has a cost • GCR = 1 might not be profitable, because Pnom has some cost and Egrid some different revenue Also economical aspects decide where the optimal solution lies
- 11. Page 11Page 11 Basic Economic Analysis • Simplified Financial analysis: Balance = Revenues - Costs • The most profitable scenario is in between the extremes GCR = 0 or 1 Pnom Area Investment 1500 $ / kWp 8 $ / m2 O&M 29 $ / kWp yr 0.03 $ / m2 yr Return 0.13 $ / kWh Timespan 16 years fixed Pnom fixed area Timespan is not necessarily the system lifetime
- 12. Page 12Page 12 Profitability as function of time • The best system design can be a function of time horizon • Optimizing short term returns neglects future benefits • Very sensitive to financial input variables • This kind of analysis helps to get a feeling for the sensitivity to different variables 12 years 14 years 16 years 18 years Fixed area scenario
- 13. Page 13Page 13 More complex economical analysis • Levelized Cost of Energy (LCOE) • Discounted Payback Period (DPB) 𝐿𝐶𝑂𝐸 = 𝐶 𝑛 1 + 𝑑 𝑛 𝑁 𝑛=0 ÷ 𝑄 𝑛 1 + 𝑑 𝑛 𝑁 𝑛=1 Cn : Costs in year n Qn : Energy output / saving in year n d : discount rate ∆𝐼 𝑛 1 + 𝑑 𝑛 𝐷𝑃𝐵 𝑛=0 ≤ ∆𝑆 𝑛 1 + 𝑑 𝑛 𝐷𝑃𝐵 𝑛=1 DIn : Incremental investment costs DSn : Annual savings net of future annual costs d : discount rate • IRR, NPV, etc… * W. Short, D.J. Packey, T. Holt, ‘A Manual for Economic Evaluation of Energy Efficiency and Renewable Energy Technologies’, March 1995, NREL/TP-462-5173 * *
- 14. Page 14Page 14 Boundary conditions • Boundary conditions help to zero in on optimal solution • For example: − Clearance between sheds − Maximum / Minimum EGrid − Maximum payback period − etc. • It can also help to identify weaknesses (like losses due to clearance, sizing too close to limits, etc.) fixed Pnom fixed area
- 15. Page 15Page 15 Net Metering Load peaking at noon, Constant over the year Constant self-consumption favors winter layout • Best solution depends on price ratio of saved and sold energy summer layout winter layout
- 16. Page 16Page 16 More Examples • Any figure that can be expressed as function of the design space, Pnom, area and the output variables, is a potential candidate for an optimization plot Life Cycle Emissions Pnom Area Construction 150 kgCO2 / kWp 80 kg CO2 / m2 O&M 100 g CO2 / kWp yr 3 gCO2 / m2 yr Avoided 0.5 kgCO2 / kWh Timespan 16 years
- 17. Page 17Page 17 fixed area Summary • Batch simulations allow systematic variation of design parameters • For large installations we assume scalability of variables • Optimal configuration can quickly be found • Scenario can be adapted (fixed area vs. fixed Pnom) • Figures of merit give a measure for optimization • Boundary conditions constrain design space and help to identify the optimal solution fixed Pnom This optimization is a guide towards the best design, it does not replace a detailed simulation of the final design choice
- 18. Page 18Page 18 Outlook Further analysis − Additional economic measures − Superimposing of plots − Simulation with variable grid tariffs − Study variable E-W orientation Implementation in PVsyst • Add more batch parameters and output variables − Number of sheds − Consider also tracking devices − Output variables of financial evaluation • Simplify the use of batch simulations − Automatic generation of batch parameter files − Parallel processing • Integrate visualization of batch results into PVsyst