Excerpt of a presentation from the EAI Solar PV Developer – EPC Meet, Chennai, Jan 22, 2013
| Mr. Amit Kaushik, Director of Energy Analytics & India Renewables Operations Centre of MEMC, spoke on EPC/Vendor Quality issues impacting Plant Performance. During his presentation he discussed the features and limitations of various metrics used in evaluating solar plants, variability present in such measurements, and the reasons for various failures in power plants |
Mr. Amit Kaushik commenced his presentation by speaking of SunEdison/MEMC’s Renewable Operations Centre in Chennai which remotely manages 1GW of installations around the world, which gives him access to a lot of data and analytics to draw upon. He mentioned basic metrics that every power plant operator needs to know
- Energy Production – Most basic, how much energy does the plant produce
- Insolation – Sunlight received from the plant
- Yield – Energy produced per unit of plant capacity
- Performance Ratio – Proportion of received sunlight converted into electricity. This is an important ratio that effectively measures every feature in the design and implementation of the plant
- Investment Performance Ratio – Proportion of actual energy produced to reference energy produced. Reference (or expected) energy produced is calculated from simulations based on boundary conditions such as land, design, and weather conditions
- Operating Performance Ratio (OPR) – Proportion of actual energy produced to expected energy produced with actual weather. This formula is used to overcome the limitations of the Investment Performance Ratio that has an assumption of weather conditions built into it which might not coincide with the actual weather experienced at the site. The two most important weather factors that determine energy produced are insolation and temperature which are measured to calculate the total energy that could have been produced at the site if the plant is functioning as it should
- Weather Performance Ratio – Proportion of actual insolation to reference insolation. This ratio explains the variation from financial forecasts that can occur due to local weather (weather risk)
- Availability – Proportion of sunlight hours where the plant was ready to produce electricity (measured as weighted average of inverter availability). The limitation of this calculation is that it gives equal weight to any hour of the day that the inverter is not generating electricity even though in practise different hours of the day impose a different loss due to non availability
Mr. Amit Kaushik now shared a sample chart plotting OPR for two months after all days with plant issues had been removed from the data. He pointed out that despite being normalised for weather the data still showed a degree of variation that cannot be explained. He emphasised that if the plant underwent performance evaluation for facility acceptance on a day when the variation dipped below the projected output it would fail the evaluation. Therefore the pass/fail criteria for facility acceptance should be framed keeping in mind such variation, and allowing a suitable number of days (7-10) for measuring output to ascertain indicative performance.
He next shared a chart that showed OPR compared with % of the plant that had been cleaned (15-17 day cleaning cycle). It showed that maximum output was reached when half the plant had been cleaned, indicating that cleaned panels became soiled before the cleaning cycle was complete. The rate of soiling is dependent on local conditions, and can be contained by increasing manpower to shorten the cleaning cycle. Facility acceptance should be performed when the plant is at maximum output during the cleaning cycle.
On quality issues, he said that 4% of energy was lost in 2012 due to various issues on 50 MW of installed capacity. This could be attributed to
Our specialty focus areas include 
- Construction – 35%
- Use of under-speced parts – Load imposed is greater than they can withstand
- Poor workmanship
- Construction issues usually manifest themselves within the first quarter of the plant being commissioned, and then gradually reduce as they get resolved
- External grid – 27%
- Grid fluctuations
- Planned and unplanned grid shutdown
- Parts/Materials – 24%
- Modules and inverters related
- While failure of parts is not unexpected as no single part can last forever, the extent of loss attributed to such failures suggests that they are more frequent than allowed for
- Other – 13%
He concluded his presentation by discussing supplier quality issues where he touched upon
- Potential Induced Degradation (PID effect) in modules (leakage of current into the frame of the modules. He mentioned that 2012 saw a sudden spurt in PID related failures compared with previous years, suggesting that price wars were taking their toll on quality
- Causes >50% reduction in module output
- Conditions required for PID
- – ve floating
- Crystalline modules
- High voltage of strings (> 600 V)
- Humidity
- Site temperature
- Cable faults
- Excessive hardware and firmware related failures in inverters
- Inability to absorb grid fluctuations led to IGBT failures
- Hardware and firmware level modifications are required to reduce failure rates
