Conference Presentation: Quantifying Sources of Mismatch


Published on

Slides from a presentation given at Solar Power Generation 2013 highlighting the benefits of panel level optimization as related to panel to panel mismatch in solar arrays.

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • Hello Everyone, my name is Evan Sarkisian, and I am the marketing director for Tigo Energy. This presentation is a myth-buster. We will be quantifying the sources of mismatch in solar arrays to try to up-end the illusion that all panels operate uniformly at standard test conditions and will age at exactly the same rate for the life of the system. Normally, my presentation covers all the benefits of module-level optimization [next slide]
  • But today I’m just going to focus on this one [next slide]
  • Increasing energy harvest. And I’m actually going to focus on why centralized systems lose power from different types of mismatch. But let’s start with a quick introduction of our technology [next slide]
  • In traditional arrays, the string current is defined by the weakest module. When one panel underperforms, the whole string is affected similar to Christmas lights. There are a few ways to solve this problem, but impedance matching is the technology that sets Tigo Energyapart. It is the most efficient way to mitigate mismatch, because it only operates when necessary. Traditional solutions boost or transform power constantly, even when they are performing perfectly. Smart modules powered by Tigo Energy log less working hours per kWh and use a fraction of the part count of alternative approaches.
  • Tigo Energy is a critical part of the world’s first TUV certified smart module, which embeds all the benefits of panel level optimizers and micro inverters directly into the module. We’ve partnered with leading module manufacturers to make the installation of intelligent panels easier than ever before.
  • There are now millions of optimizers in the field reporting data at the module-level. We are measuring every 2-seconds with an accuracy better than 1%. The industry has never had more data to answer the question of why is there so much mismatch between panels in “perfect” arrays. Here we see a young system experiencing a voltage variance of 15%. We’ll be looking at all the major sources of mismatch starting with differences that are inherent in the panels [next slide]
  • If modules receive the same irradiance you would still see variance in performance just because of the manufacturing process. PV cells are crystalline structures, like snow flakes, and no two solar cells are ever identical. Impurities in the air, residue buildup in tooling, and thermal drift all contribute to variance in modules. Manufacturers are continuously working to improve their binning range, and you can see in the table here that the average range is well above 5%. But beyond the stated numbers we need to consider the accuracy of the flash testing done to get these ranges. These are typically accurate to plus or minus one percent, further increasing the actual range of mismatch in module performance. In addition to module variance out of the factory, there is also variance in how fast modules degrade [next slide]
  • Module degradation is well understood, what is less commonly known is the variable nature of degradation. Not all modules will age at the same rate, and this leads to increased mismatch from burn-in as well as over the long-term. NREL completed a study of module degradation rates and found a wide difference in degradation between modules. A significant number degraded between 0 and 1 percent per year, but a 1 percent variation per year will lead to a 5 percent difference in power production throughout the entire array by year 5. A smaller, but not trivial, amount of modules showed significant degradation rates ranging from 1 to 4 percent per year. Within 5 to 10 years these modules will be underperforming their peers by as much as 20 to 40 percent. All the variance I’ve described so far is inherent to the module, and would be an issue even if you were designing in a lab. But since no solar array will experience laboratory conditions, let’s take a look at some environmental factors that impact mismatch on real world installations. [next slide]
  • Here we see the difference in irradiance a site will see just from clouds passing over head. This data set was collected by NREL and it gave us our first look into second level data of the effect of clouds. NREL used 17 different light meters and the annual average deviation was 15.4 percent, this is just from clouds. That’s a 5-8 percent energy loss due to variable light levels. On top of that, modules will get dirty affecting production beyond irradiance differences. [next slide]
  • Dust buildup alone can reduce standard module performance by 30 watts, and if part of the array is closer to trees or roadways the effects of soiling will be felt unevenly throughout the array. A proximity to birds adds a harder to predict, but equally impactful, source of mismatch to arrays. And since soiling effects require a cleaning visit to fix, they are often left unattended for 3 to 6 months at a time. This means that some modules will operate for an extended time at a reduced efficiency, creating a hotspot of increased resistive losses, which will accelerate degradation for that module. Based on historical data, most developers use 1 to 4 percent mismatch loss due to soiling.
  • Thermal mismatch affects module performance relative to the temperature coefficient of the module. This is typically modeled at 1 percent loss for every 2 degrees of temperature increase. Within Tigo Energy’s fleet we’ve measured an average variance of 4 to 7 degrees celsius, with 20 degree spreads commonly found throughout the fleet. This equates to an average mismatch loss of 2 to 4 percent. We’ve had this validated with a presentation by First Solar that mentioned a 10 percent average module temperature difference throughout their ground mount arrays. So far all the variance I’ve described has assumed that 100% of the modules will work perfectly for the life of the system. Solar modules are highly reliable products, but they are not immune to failure. [next slide]
  • Half a percent of modules have a failed bypass diode the day they are installed. Failed diodes are almost impossible to detect with traditional measuring devices since a load must be drawn for the issue to manifest. Which means, one third of a module’s power can be entirely missing for the life of a system. Working bypass diodes can operate at temperatures 2 times ambient, leading to high stress operating conditions. And as I am about to show you, aging systems continue to suffer from accumulated wear and tear
  • This is a case study involving a 9 year old retrofit system in Germany. Module-level optimizers were installed to diagnose and isolate system issues. Immediately after installing, the system experienced a 25% increase in production just from module-level power point tracking. With module-level monitoring the system owner quickly identified 10 modules producing at 90% relative to neighboring modules, 12 modules at 50%, and 76 modules producing at 20%. In addition, 2 entire strings were experiencing open-circuit faults. The point being, solar systems are exposed to extreme environments and aging is inevitable. These things are in the sun all day, and temperature cycle from hot to cold each night. Rubber, polymer, glass and circuit connections are all exposed to thermal expansion and humid conditions. That covers all the main sources of mismatch, now I’d like to show some real-world data [next slide]
  • Here is the standard deviation range for over 10,000 systems using Tigo Energy module-level monitoring, with hundreds of commercial sites included. This represents over 60 gigawatt hours of production. We’re reading power, current and voltage every two seconds with accuracy greater than 1 percent. Across the entire range we see an average standard deviation of 11.3 percent. But these numbers are a little skewed. Tigo Energy is a growing company. We doubled our installations last year, which means one half of our systems are less than a year old. Let’s look at an older data set. I just found this next study, and the information is eye-opening. [next slide]
  • Arizona State University completed a study of 1900 modules using Arizona Public Services’ STAR research facility. These modules were cleaned and an IV curve trace was run to measure degradation. They found a degradation rate well above the 0.5% degradation rate used by many models, but more importantly, they found a significant mismatch in the rates of degradation.
  • Thank you all for listening. I hope you are now mismatch experts that can challenge anyone that hides behind a spec sheet and argues that mismatch isn’t an issue. Please contact us if you’re interested in learning more about some of the other optimizer values I didn’t cover, and feel free to raise your hand now if you have any questions.
  • Conference Presentation: Quantifying Sources of Mismatch

    1. 1. Quantifying Mismatch in Solar Arrays February 2013 2.2 MW Installation Talmage Solar Engineering
    2. 2. Smart Modules Deliver: • Arc, Fire and Safety Hazard Detection • 30% Longer Strings, Lower BOS Costs • Reduced Customer Acquisition Costs • Increased Energy Harvest and System Uptime for Proven ROI • Module-Level Monitoring, Reduced O&M and Commissioning Costs • 25% More Power Density / Efficiency
    3. 3. Smart Modules Deliver: • Arc, Fire and Safety Hazard Detection • 30% Longer Strings, Lower BOS Costs • Reduced Customer Acquisition Costs • Increased Energy Harvest and System Uptime for Proven ROI • Module-Level Monitoring, Reduced O&M and Commissioning Costs • 25% More Power Density / Efficiency
    4. 4. Impedance Matching Traditional Modules – String Current Defined by Weakest Module Smart Modules – String Current Defined by Strongest Module • Each Module Produces 100% of its Capacity Strong Innovation • 50 Patents Submitted • 11 Patents Granted
    5. 5. The World’s First Certified Smart Module • Junction box diodes replaced with intelligence • The only product certified by TÜV for North America and Europe • 3rd party validation
    6. 6. Mismatch in “Perfect” Arrays • Voltage Variance of 15% on Perfectly Architected Array • < 2-Years-Old Measured at Peak Production Time
    7. 7. Manufacturing Mismatch Bin Power Range # of MFG’s surveyed 3% total 2 5% total 3 6% total 3 10% total 2 Mean 5.9% range Median 5.5% range • Crystalline structures have inherent differences • No two solar cells are ever identical • Typical binning range is over 5% • Flash testing accuracy broadens range
    8. 8. Variable Degradation • Module degradation is not uniform • Approximately 0.75% increase in standard deviation each year • Data published by NREL shows differences in degradation speeds across modules • In year 5, this represents 1-2% additional losses, which grow to ~4% in year 10 and 12% in year 20 Source: “Outdoor PV Degradation Comparison”. D.C. Jordan, R.M. Smith, C.R. Osterwald, E. Gelak, and S.R. Kurtz. NREL Conference Paper CP-5200-47704 PV Evolution Labs Tigo Energy Analysis
    9. 9. Mismatch from Clouds • NREL Oahu dataset: – 17 lightmeters at one location – 1-second intervals – 1 full year of data • 15.4% average standard deviation¹ in irradiance due to clouds • Corresponds to 5-8% in lost energy due to mismatch Note¹: Weighted by power production at each second; excluding lightmeter readings lower than 75W/m² Source: Tigo Energy analysis 0% 5% 10% 15% 20% 25% Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec Monthly Standard Deviation
    10. 10. Uneven Soiling • Driven by various factors: • Level of dust in the air • Proximity to roads • Proximity to birds and plants • Long term variance leads to hotspot • Reduced efficiency = increased resistive losses = further reduction of efficiency • Estimated mismatch losses: 1-4%
    11. 11. Thermal Mismatch • Measured thermal mismatch on flat commercial rooftops in Northern California: – Average of 4-7 C range between max & min – Corresponds to 2-4% mismatch in module power output – Maximum of 20 C spread • First Solar said they have module operating temperatures differ by 10 C or more¹ in ground-mount arrays Source¹: “PV Performance Modeling Workshop Summary Report”; Cameron, Stein, and Tasca; Sandia National Labs; May 2011 Source: Tigo Energy analysis 68 71 71 72 72 74 75 73 73 70 71 72 68 69 67 74 77 79 78 77 78 78 75 70 70 71 72 71 74 76 79 85 81 82 75 77 73 73 70 67 68 74 77 78 79 77 75 77 80 74 69 68 65 68 71 71 74 75 72 72 74 72 72 72 67 Module Temperatures on a Rectangular Array
    12. 12. Failed Bypass Diodes • 0.5% of modules have failed bypass diodes on day one. • Results in 0.5% system power loss • Difficult to detect with typical measuring devices
    13. 13. Accumulated Wear and Tear • 9 year old retrofit • 25% improvement in performance when optimized – more when modules were replaced • 10 modules (3.0% of system): producing 90% of peers • 12 modules (3.6% of system): producing 50% of peers • 76 modules (22.7% of system): producing 20% of peers • Two strings (7.2% of system): open-circuit fault
    14. 14. Mismatch Summary Module binning mismatch 3.0% Thermal mismatch 1.0% Variable soiling 0.5% Cloud variance 1.0% % failed diodes found 0.5% Variable degradation (per year) 0.5% Total Mismatch (Year 1) 5-7% Total Mismatch (Year 5) 7-9% Total Mismatch (Year 15) 12-15%
    15. 15. Standard Deviation Range from 1 to 25%+ In Tigo Energy’s installations, average standard deviation of 11.3%
    16. 16. ASU-APS (STAR) Test Results Source: Dr. Mani, PV Power Plants Conference, 12/2011 • 1900 modules cleaned prior to inspection • 10-17 years in the field • Annual degradation range 0.93 – 1.92% • Modules sold at ~$4/W 0.00% 5.00% 10.00% 15.00% 20.00% 25.00% 30.00% 35.00% 10.7 12.3 17 Weakest Strongest Degradation Comparison (Strong vs. Weak)
    17. 17. Thank You! Evan Sarkisian Tigo Energy