How important is the accuracy of the PV generation forecast?

1. Foreword

Starting from this year, the distribution of photovoltaic project scale indicators will adopt the "competitive allocation method". From several provinces that have issued competitive allocation methods, the score of the feed-in tariff is between 24 and 30 points.

In order to obtain the project scale index, we must compete, and other conditions are equal, it is difficult to open the gap, feed-in electricity price is the most important indicator to open the gap. How to get the scale index with the most reasonable feed-in tariff? In this case, investors must calculate the back-calculated tariff according to the target project rate of return to bid.

Not long ago, the bidding price for Anhui's 2016 photovoltaic project scale index bidding was between 0.945 and 0.98 yuan/kWh (the local benchmark price was 0.98 yuan/kWh). The scale of the project bid at 0.945 yuan/kWh electricity price has exceeded the local pre-allocated scale.

0.945 yuan/kWh is the back-calculated electricity price obtained by the investor based on the target rate of return and other boundary conditions. However, what are the factors that affect the back calculation of electricity price? Figure 1.

At present, China's photovoltaic has been connected to the grid more than 60GW, we have an accurate grasp of the project investment level and development trend. However, the prediction of power generation has great deviation and poor accuracy.

The accuracy of power generation prediction mainly depends on two factors: the accuracy of solar energy resource data and the accuracy of system efficiency.

2. the accuracy of solar resource data

As a professional technician of resource analysis, I feel that domestic photovoltaic power plant investors generally despise "solar energy resource analysis" and are unwilling to spend money on accurate data of weather stations.

1 Is satellite data reliable?

Because investors are reluctant to spend money on weather station data, technicians often use free satellite data for analysis. However, the accuracy of free satellite data is a concern. The following figure is a group made by the Wind Energy and Solar Energy Research Center of the China Meteorological Administration: NASA data compared with weather station observation data.

Is the measured data of the project site reliable?

Some investors said that there is a production project next to my project site. We have a good relationship. Let's use their data, or time by time!

I have seen a lot of field measured data, the accuracy is also worrying. The main causes of error are:

1) Buy cheap observation instruments, the accuracy is not high, and they have not been professionally calibrated before installation;

2) Observation instruments are installed in unreasonable positions, and the surrounding environment is greatly affected;

3) Daily lack of professional maintenance.

Therefore, the results of field measured data are sometimes larger than the deviation of satellite data!

In addition to accuracy, the length of time is also an important factor affecting the accuracy of the forecast. Our photovoltaic power plant is to operate for 25 years, but 25 years, the total annual radiation of solar energy resources may change greatly. The following figure shows the change curve of the total annual solar radiation in a certain place in 30 years. The difference between the maximum value and the minimum value is 11%, and the average deviation is 2.8

If the observation year happens to be the best years for solar energy resources, the prediction of power generation will be too large. Personally, I think it is difficult to accurately predict the future power generation with less than 10 years of historical data.

3 What is the impact of power generation errors?

How much impact will the power generation error caused by inaccurate resource data have on the project revenue? The following figure shows the sensitivity analysis of the revenue rate of a project.

It can be seen that power generation is undoubtedly the factor that has the greatest impact on profitability. The effect of changes in different factors on the cost of electricity was also calculated before, and the results are shown in Figure 4.

In summary, if the use of NASA data results in an error of-6% to 20% in power generation forecasts, or if the use of data for a given year results in a prediction error of more than 5%, then the expected project returns of investors will also vary greatly.

Revenue from electricity sales = electricity generation x electricity price

If the power generation is overestimated by 10%, the reverse electricity price will be underestimated by 9.1.

That is, the project would have been able to earn the expected return of $0.98/kWh, but because of the overestimation of power generation, the expected return would have been calculated at $0.89/kWh. If the investor bids for development rights at $0.89/kWh in order to obtain the indicator, the future project yield will be lower than expected.

Compared with the tens of millions of electricity revenue in the future, spending tens of thousands of dollars to purchase accurate resource data during the preliminary work of the project is simply a drop in the barrel!

Accuracy of 3. system efficiency

There are many factors affecting system efficiency, especially affected by the level of operation and maintenance in the later stage, no one can accurately calculate the accurate system efficiency of a certain place or project.

The best way to solve this problem is to judge by the big data of monitoring software. At present, there are many third-party monitoring platforms and optical power prediction platforms in China.

By analyzing the big data accumulated by these platforms, the probability distribution of system efficiency in a certain area can be obtained, so as to obtain the system efficiency with the highest probability of a certain type of power station in this area.

The system efficiency of our power station is difficult to reach the highest value, and it is also difficult to reach the lowest value. The most likely one is the one with the highest probability. No one escapes probability.

Therefore, some big data platforms for monitoring and power prediction that are being built now are of great significance to accurately predict power generation in the future.

4. epilogue

In order to get the scale index with the most reasonable feed-in tariff in the "competitive allocation", it is necessary to accurately predict the power generation.

Therefore, in our preliminary work, we must pay attention to the accuracy of solar energy resource data; at the same time, we also expect the big data platform of monitoring and power prediction to provide more useful statistical data analysis results for the industry.