Discussion on Simulation Accuracy

[xueting]

Hello,

I have recently encountered a problem with PVsol Premium and I hope it can be answered. I set up an experimental bureau to compare the simulated power generation with the actual power generation.First, I imported the data processing of the environmental irradiator, which recorded temperature and irradiance , into PVsol Premium,Then, according to the actual scene modeling and simulation, the final result shows that the simulated power generation under strong irradiation is higher than the actual power generation.The red line is the simulated value, and the blue line is the actual power generation value.In simulation I set 4% dust loss.I want to know why the deviation is so large, and secondly, what parameters can be input in order to improve the accuracy during simulation?

[developer_mh]

Dear xueting,

simulation comparisons are really interesting, I am happy to see that you want to test PV*SOL simulation values against measurement. When looking at the details, there can be a lot of reasons why the results differ. In order to answer your question we would have to know a lot more about the two sides of the experiment:

First, the measurement: How did you measure the global irradiance? Which sensor did you use? In what environment was the sensor installed? In which intervals was it cleaned? Did you measure global horizontal irradiance or tilted, i.e. in the PV module plane? How many modules do you have on your test stand? How are they connected? Do you measure the electricity on the DC or the AC side?

Second, the simulation side: How did you determine the input parameters for the simulation? For example, the module data: Did you use data from our database or do you have flasher data that you were able to enter? Which models did you choose to calculate the irradiance on the tilted plane?

There are really a lot of factors that influence the outcome of such a comparison. Without a lot more details on the measurement setup and the PV*SOL project file, however, we can't answer your question in a satisfying manner.

Hope that helps, kind regars,

Martin 

[xueting]

Hello, for the global irradiance, the environmental detector is used to obtain the global irradiance, and the SN model of the device is EM00101990088966.The environmental detector was placed horizontally and installed about 3 meters away from the experimental components.As far as I know, the device is basically not cleaned.The environmental detector is placed horizontally, while the PV modules are mounted on brackets with a 17° inclination. In the simulation model, the components are also modeled with an inclination of 17°.In the comparison process, 12 modules were used, and every 6 modules were connected in series and then connected to the inverter in parallel, and the power generation on the AC side was compared.

The photovoltaic module is CdTe, and its STC performance data is newly created in the software according to the manufacturer's data.
The detailed model parameters are as follows：

 

[developer_mh]

Dear xueting,

this looks good so far. So as a next step I would recommend to look at the irradiance values before and after the simulation. Best option here would be the diagram editor where you can select the time series you want to analyze. You can select the "Irradiance onto horizontal plane" and the "Irradiance onto tilted plane" and see how it looks like. You can right click and copy the values to the clipboard, and compare them to the values that you imported. Just to be sure, also have a look at the elevation of the sun. Its curve must be in sync with the irradiance values, that is, sunrise and sunset must not be shifted.

Next check would be the PV module data that you entered. Especially the low light behaviour might be a reason for an unusually good performance.

If you wish, you can also send your irradiance data as well as the project file so that I can have a quick look. you can send the files here in the forum via private message.

Kind regards,

Martin

[Jimmy]

Hi,

I have a few questions regarding this. I've never input climate data myself so there might be a few things that I don't understand about this. But anyway.

A few things stick out to me as worth mentioning/questioning.

1. The simulation is done in 1 hour steps while I assume that the data is in minute steps.

2. If precise radiation values are recorded and used, should the albedo really be set to 20%? Seems like the climate data has already accounted for albedo and that we're adding another 20% on top of actual values.

I'd also be very interested to know the precise setup of the system. What panels, inverter and cables are being used. How old are the panels and have they been checked that they actually produce their nameplate under STC?

Lastly I want to mention the possibility about shading of the system being larger than the shading of the sensor and how that might affect the simulation since I assume that shading environment and horizon were not 100% accurate (if modeled at all). 

Just a few thoughts. 

[xueting]

hello，Jimmy

1.Both the imported irradiance data and the simulation are in 1h steps.

2.As for the second question, I don't understand either

3.The photovoltaic module used in the experiment is the module of Longyan company model ASP-S1-105, the module is not in the software library, so I imported it according to the specification provided by the manufacturer (performance in STC: 1000W/m², 25°C , AM1.5).The inverter adopts Huawei products, model: sun2000-10KTL-M1.The total cable loss set in the software is 0.4%, and there is no specific distinction between what material the cable is made of.The service life of the components is linearly attenuated: 80% of the output is still available after 25 years.

4.The environmental monitor is not blocked by shadows, and the components are blocked to some extent, which is also considered in the simulation model.

I'm sorry that due to company security restrictions, I can't send the project files.

[Jimmy]

Hmm... I'm thinking that there has to be some fundamental flaw in how this project is setup since all of my installations are pretty close to accurate when comparing simulation and reality. And since the biggest new factor is imported climate data it should be something with this.

When you check the PV System Energy balance, is it the same global radiation at the module as your measurements?

 

[xueting]

23 hours ago, developer_mh said:

Dear xueting,

this looks good so far. So as a next step I would recommend to look at the irradiance values before and after the simulation. Best option here would be the diagram editor where you can select the time series you want to analyze. You can select the "Irradiance onto horizontal plane" and the "Irradiance onto tilted plane" and see how it looks like. You can right click and copy the values to the clipboard, and compare them to the values that you imported. Just to be sure, also have a look at the elevation of the sun. Its curve must be in sync with the irradiance values, that is, sunrise and sunset must not be shifted.

Next check would be the PV module data that you entered. Especially the low light behaviour might be a reason for an unusually good performance.

If you wish, you can also send your irradiance data as well as the project file so that I can have a quick look. you can send the files here in the forum via private message.

Kind regards,

Martin

Dear Martin

I'm sorry that due to company security restrictions, I can't send the project files.

Based on your comments, I compared "Irradiance onto horizontal plane" , the "Irradiance onto tilted plane" and imported irradiance data. The "Irradiance onto horizontal plane" is basically the same as the imported irradiance data.The "Irradiance onto tilted plane" data will be higher than "Irradiance onto tilted plane", but I think this is reasonable, because I am in Dongguan, China, the module is installed with a south angle, the "Irradiance onto tilted plane" data Because the inclination angle will increase.
As for the sun height curve and the irradiance curve are basically synchronized.

As for PV module data, the manufacturer only provides STC data (performance in STC: 1000W/m², 25°C , AM1.5).Low light performance parameters not provided.From 10:00 am to 3:00 pm, the irradiance value is about 400-900kWh/m2. Will this value be considered by the software as a low light scene?

In addition, I would like to ask, during the simulation, if I select the meteorological data of a certain place in the software library, do I still need to manually input the sun altitude?

[developer_mh]

Hi xueting,

I understand that it is not possible for you to send the file. It will be difficult to identify the problems then, I guess.

On 4/8/2022 at 9:43 AM, xueting said:

The "Irradiance onto tilted plane" data will be higher than "Irradiance onto tilted plane", but I think this is reasonable

Yes, this is absolutely reasonable.

On 4/8/2022 at 9:43 AM, xueting said:

As for the sun height curve and the irradiance curve are basically synchronized.

This is good. Additionally, can you check if the peak of the daily sun height curve is in sync with the peak of the daily irradiation curve?

 

On 4/8/2022 at 9:43 AM, xueting said:

do I still need to manually input the sun altitude?

No, the position of the sun is calculated in the software, for each time step.

 

Some more question/issues to look at:

• Can you post screenshots of the PV module database entry?
• Do you have the measured data on the DC side of the inverter?
• Did you measure and import one complete year of irradiance data? From 01. January to 31. December?

 

Kind regards,

Martin

[xueting]

On 4/12/2022 at 3:35 PM, developer_mh said:

Can you post screenshots of the PV module database entry?

hi，Martin，

The screenshot is shown below：

On 4/12/2022 at 3:35 PM, developer_mh said:

Do you have the measured data on the DC side of the inverter?

I'm sorry, I can't measure the data on the DC side of the inverter.

On 4/12/2022 at 3:35 PM, developer_mh said:

Did you measure and import one complete year of irradiance data? From 01. January to 31. December?

As for the weather data, I only imported the data from December to March, and replaced the rest with zeros.