Michele Oliosi

Hi, It seems excel doesn't read your entries as numbers. Usually numbers are right-justified, i.e., I suspect most entries are read as text or some other type of cell. Please check the cell type, and make sure that all values are actual numbers. Regarding the negative values, it is due to night auxiliary consumption, either from your “Detailed losses” > “Auxiliary losses” definition, or from the inverter definition, or from the transformer night iron losses.

Thanks @dtarin this is a great answer !

No, if the module is marked as “low-light efficiency seems over-evaluated”, it means that the model behaves in an optimistic way in low-light conditions. The message serves to point out that it is “not realistic”.

Dear Dimitri, Thank you for your question. Just a quick reminder, the official language of the forum is English, so you will benefit most from it when asking questions in English (there is no automatic translation). ------- One (difficult to configure) option in your case would be to use monofacial modules of different efficiencies to model the front and the back of the bifacial module. These should be connected in parallel to add up the currents. ------- Another option is to assimilate the bifacial gain to a simpler variant with only a single orientation. Choose only one side of the cross and run the bifacial model. Running the simulation with the backside irradiance modeling on and off will allow you to extract a relative bifacial gain. The perpendicular second side of the cross should be modeled as shading objects (rectangles). Since the bifacial model is not counting the perpendicular side of the cross modeled as shading objects, the shading factor for the backside should be increased accordingly. The proportion of backside shading is likely around 25%. To get a precise idea of the backside shading factor to be applied, you can observe the shading factor for the front side. Then, apply the same shading factor to the backside by means of the parameter in the bifacial model. The mismatch parameter for the backside should also be assimilated to the electrical shading loss of the front side.

Indeed, this hasn't been fixed yet. It is currently planned for version 8.1. There is one workaround I can think of: if the weather data has been imported from a file, you can also import it separately month by month as different files (by splitting the original file by months). This will give you MET files that contain only one month. Simulating with them will limit the simulation to the date range included in these files.

Hello, In the current PVsyst version, there can be only a single tracker orientation at a time, unfortunately. Therefore, splitting in two orientations is indeed the only way. You could average them in a single orientation, but this won't be a good approximation.

Thanks for the feedback. We have a ticket to improve this in version 8. Indeed, the bifacial model window was initially aimed at experimenting with parameters, but it is quite confusing when designing projects.

No, since PVsyst determines the efficiency normalized on the module area, it is necessary to work with the latter for consistency. There is no difference in terms of accuracy, it is more a matter of convention.

Not in version 7, trackers are not compatible with fixed orientations. You need to simulate them separately. In future version 8, this will however indeed be possible.

Indeed, thanks for the answer ! 7.4.6 is correct, i.e., a 2.3 cm imprecision will generally cause shadings shorter than a cell width, by which electrical shading losses are not so critical.

This message is not very explicit, and will be improved in later versions. Basically, either: You use the orientation "unlimited sheds" and put "No shadings" in "Near shadings". The shading loss is already included in the "unlimited sheds" model. You use the orientation "fixed plane" and use a shading model in "Near shadings"

As mentioned, in PVsyst you can only have the power output for a system as a whole. If you need the power output per MPP or per inverter, you should define a system containing only 1 MPP input, or only 1 inverter.

Hi, I don't think that makes sense, in particular for P90. A percentile is a statistical summary value for a given distribution of results. Generating an instance of time-series results that gives a particular summary does not have any significance, i.e., you cannot learn much from this. We can gladly discuss details if you do not agree. The P50 is a bit special because it tells you the average results, and you can assume that aiming for a given P50 may model a “realistic time-series”. However, I am not sure what you can gain from having a P50 time series shown in a table in a report. Analyzing the time series can give you some statistical information on the distribution of results, but you would need some summary values to make sense of that. At the moment, the only way to get a time series to be shown in the report (but only as a list of yearly values, not monthly) is to use the aging tool with previously prepared time series weather data. You can also generate monthly output files from running the time series in the aging tool and make your own formalized output from that.

Dear Shivam, Can you send your project to support@pvsyst.com? We will need to analyze it in details, since it changes only for one of your sub-arrays. Once you send it we will look into it asap.

Hi, Unfortunately, PVsyst can only export variables for the system as a whole. In order to have power losses per MPPT, you need to make different variants with one of the MPPTs only and run the simulation for each of them. This will be very time-consuming, so I wouldn't advise doing it. Same thing for the shade by module, it is not an export value for PVsyst. You need to make a special scene with a single module to get the shading values for a single module. At the moment, in PVsyst, the best in terms of stringing is the single line diagram. However, there is no spatial layout associated with it. You can also print the "Module layout" definitions to color-code the stringing with a spatial layout. However, there is no legend associated to it.

Yes that seems correct !

Could you send the shading scene at support@pvsyst.com ? We could take a look at it. Indeed, 7% electrical shadings seems like too much. I also don't understand why it's 99% shadings and not 100%. Also are you modeling with which version and which shading model (fast or slow)?

This is not simple, but let's see what we can do. First, mutual shadings tend to happen starting with 20°- 30° angles of sun height. These are the times when backtracking is interesting, but also where it seems you are expecting glare. So you have a double constraint on these low sun angles. Note that at times when backtracking is limited, you will have mutual shading. In these cases, it will be best to use just regular tracking. This is because backtracking is only convenient whenever you are removing most of the mutual shading (and electrical mismatch effects). So once you are in a mutual shading regime, I think it is probable that regular tracking will be best. Based on that, you can run two simulations: one with backtracking and one without. For each of them, you can generate an hourly output file. Then you can process the data, and use the regular tracking one whenever the backtracking rotation angle would be < 20°. The resulting dataset can then be summarized to obtain yearly yields, etc.

This error message was meant to be present in the previous versions of PVsyst but a bug prevented it from appearing. This has been fixed in the latest versions of PVsyst, that’s why you see the error message now. To get rid of the error message, you can go to the Home window > Settings > Edit advanced parameters, search for “spread” and modify the following value: I hope this helps. Note that cases which have a very large tilt spread may induce a certain uncertainty in the transposition calculation.

Since both front and back-face energy contribute to the module conditions (irradiance and temperature) it is not possible to split them. The simulation is not simply additive of back face and front face in that regard. This is shown by the loss diagram for a bifacial system: both contribute to the PV conversion:

Dear Vera, At the moment, there are no miscellaneous losses in PVsyst (it has been requested before). I would advise adding this loss to the mismatch between strings. This is conveniently in the DC side, and is consistent with part of the effect of variable terrain.

The stow position is in Phi (rotation) angle. This means that a negative angle is towards the east (as long as the axis azimuth is between -90 and 90, especially if it is close to 0).

Yes of course. You have TArray among the variables for output files: https://www.pvsyst.com/help/output_file.htm

You are comparing relative differences, but the irradiance is very low in the morning and evening. What does it look like in absolute differences ? Also, you mention a sweep of all the positions, but it is virtually impossible to have all angles. You must be doing an interpolation and sweep the positions with a certain interval? Possibly the interpolation causes some error, which becomes more important as relative values in the morning and evening due to the low irradiance. Just a hypothesis, please let us know. It is an interesting problem. What is the end objective ? A validation?

The output of the batch mode is only a CSV file, i.e., you need to use another tool to generate a histogram. The capacity of the system itself is not a parameter, but you can equivalently change the Number of strings and Number of inverters. This is equivalent to modifying the capacity of the system. If these two options do not appear in the “system parameter” tab, it means that your system has multiple sub-arrays, or some other complexity such as optimizers, etc, that prevents it to be used as a parameter. If you are unsure, please send a screenshot of your system definition, for example.