André Mermoud

This configuration is different than the previous one. Here you have defined a charging power of 100 kW, ensuring a charge in 1.6 hours under full sun. This is more reasonable. This is close to what is acceptable for Li-Ion batteries. This charging time was 10 minutes at sun in the previous case, and as PVsyst works in houly steps, this leads to some problems when simulating one full hour. Normally PVsyst should not allow a charging/discharging Power - or current - exceeding the battery data. I don't know why this did not show the correct error message. Is your battery from the database, of from an external file ? I am just on the way of correcting these extreme problems for a next version. By the way, I persist to say that implementing a battery storage for storing an energy corresponding to 8 minutes of average consumption doesn't make sense. It depends on what you want to do with your system. If it is for an UPS (i.e. ensuring the continuity of the energy supply for a little securized part of your installation), you should define the discharging power consequently.

As we should always keep a balance between the available energy and the demand, a stand-alone systems cannot work without a storage system. At each instant, the load will be satisfied by the PV production, and the battery if this is not sufficient. A possible genset is here especially for ensuring the needs when the battery is empty. Now the case where a constant power is permanently available from a genset has indeed not been foreseen in this strategy. If you really don't have a battery storage, you should consider this case as a "grid connected" system with self-consumption. The excess power generated by the PV array will be considered as a grid injection, and the energy from the Genset the energy "from grid" required by your load.

If you want to become familiar with this topic, you should define realistic conditions. Otherwise you cannot understand the problem.

Yes, thisis what I said.

I had a look on your data (sent to Support). Your data seem a bit strange. They mention a 31 June, and the 31 May is missing. I attach a corrected file, where I have suppressed the 31 June and the file may be imported. I don’t know what happened with these data. However I suggest that you shift all date values from 01/06/22 – 31/06/22 into 31/05/22 – 30/06/22.

Your battery pack (160 kWh) is completely undersized. With a PV power of 846 kWp and a max. load of 1048 kWh, it could be charged in 11 minutes, and discharged in 9 minutes. Sorry, PVsyst doesn't treat this absurd situation correctly in the present time. You should at least limit the Charging and Discharging powers (next page) to ensure a reasonable charge/discharge in about 2 hours (i.e. a charging power of the order of 160 kW/2). But such a system doesn't make sense. We have indeed to correct the model for such situations. This is on our roadmap for a very near future.

I don't understand well what you mean when you say "different efficiency profile when calculating with CEC (98.0%) and EUR (98.3)". A profile is defined as a set of values, not one only; what is 98 or 98.3 %? The EURO or CEC efficiencies are indeed a way of evaluating a yearly average of the effective efficiency, based on the expressions you gave above. This average is a weighted average of a supposed power distribution along the year, If these data powers distributions are different, the weighted result is different. The EURO efficiency is a weighted average calculated on a "reference" energy distribution corresponding to meteo data chosen as those of ISPRA (location of the JRC - Joint Research Center - at North of Italy, in the years 1990), when the CEC efficiency is based on a climate corresponding to the californian climate. Now in absence of an explicitly defined efficiency profile, PVsyst tries to reconstruct a "supposed" efficiency profile reproducing these very simplist data "Max efficiency" and "EURO efficiency" (or "CEC" efficiency). This reconstruction uses reasonable hypothesis, but cannot obviously be absolute. These hypothesis are described in the Help "Physical models used > Grid inverter > Inverter model: efficiency"

The shadings calculation should normally be based on a 3D scene. In this case you should use "Tracking, horizontal axis N/S". However for a quick pre-calculation, you can use the orientation option "Unlimited trackers", which will perform the shadings calculation according to a generic reguar system. NB: You cannot use a 3D scene with the option "Unlimited sheds".

Some answers to your questions: 1. - SOCBal is the stored energy difference between the beginning of the calculation period and the end. It cannot be higher than the energy contents of the battery of course. 2. - The SOC at the beginning of the simulation is specified in the advanced parameters, topic "Batteries". 3. - Yes, the availabe energy when discharging is EBatDisc - CL_InvB 4. - No, PVsyst doesn't treat the mode of charging the battery from the grid. This doesn't make much sense: what would be the strategy? When activating the charging ? Why?

This is not quite exact. The PR is referred to the Incident global GlobInc (result of the transposition), not the GlobEff. NB: This PR is calculated by PVsyst. You should obtain the same result. The PR indicator takes into account all the losses mentioned on the loss diagram, between "Global incident in Collector plane" and "Energy injecter into the grid".

PVsyst doesn't take an ageing of the inverter into account. On which operating parameter would it be applied ? I doubt that the efficiency could be affected, as this would increase the heating along its life. The only effect I can see is the reduction of the lifetime (MTBF). This is not addressed in PVsyst in the present time. Now for the evaluation of the produced heat during operation, the best way would be to use the inefficiency, which is a curve according to the power, and possibly to the input voltage. This is directly accessible in the simulation data (IL_Oper loss) for each operating hour.

I don't understand well what you mean. The LID loss is accounted in the simulation. It is explicitly defined in the Detailed losses parameters. The warranty mentioned on the datasheets is not registered in the PAN file, and not taken into account in PVsyst.

I don't know what you are waiting for. Our calculation model is the better way we can imagine for the evaluation of the Bifacial yield. We have deeply checked for the version 7.4.3, that the rear side incidence irradiance is equivalent to the front side (up to about 1-2%). If the bifacial results were false, this would mean that the whole simulation of PVsyst (front and rear side) is erroneous, i.e. that the transposition models themselves are incorrect. I don't know with which data in real conditions you are comparing the PVsyst simulations, and what you consider as acceptable. NB: If you have really measured data on the field, we would be happy to get them for analysing the discrepancies. I don't know how PV*SOL evaluates the vertical bifacial systems, but we don't have any reason of being quite compatible with their resuts.

The Sandia model is completely obsolete. It cannot be used without specific Sandia detailed measurements. And the latest available measurements are those included in the PVsyst database (recorede about 2005 ?). This model had been inplemented in PVsyst just for a comparison with the one-diode model of PVsyst.

In PVsyst, the only way of creating Reactive power is to set a constant specified Power factor. This cannot depend on the grid voltage, or grid frequency, as these values are not an input variable of the simulation. The grid voltage is supposed nominal, without variations. The power factor may be specified also with grid storage systems.

Probably your second screenshot corresponds to a stand-alone system, where the AC losses are not defined.

You have 244 strings on 144 MPPT. I.e. 100 MPPT with 2 strings and 44 with one string. This means 244 x 25 modules = 6'100 modules - 1 module. First option: you neglect the string with 24 modules, and define one only sub-array with 12 inverters with "Use MPPT feature" unchecked. This will give a quite reasonable result with an error of 1 /6'100 = 0.2 per mille. The result will be fully reliable in PVsyst. Second option: the problem is that the PNom sharing within an inverter cannot be used when this inverter PNom is shared with other subarrays. In this case you have to define sub-arrays with a number of strings divisible by the number of MPPT inputs. Therefore if you want something more sophisticated, you should define 4 sub-arrays: - SubArray #1 with 11 full inverters and 223 strings (uncheck the option "Use Multi-MPPT feature). I.e. you leave 21 strings for the last inverter. - SubArray #2 with 9 MPPT inputs and 18 strings - SubArray #3 with 2 MPPT inputs and 2 strings - SubArray #4 with 1 MPPT inputs and 1 string of 24 modules. and you will define an explicit PNomsharing between the subarrays #2, #3 and #4. Hopefully this should give the same result as the first option, minus 0.02%.

No, we don't have any verification. This is a theoretical hypothesis, expecially for the coherency of vertical bifacial systems. This contibution is low for other orientations (tilts around 10-20°) . It is not taken into account in the simulations without Bifacial calculation.

Yes, the optimizers may yield a little benefit, especially with irregular shading situations (building integration). With sheds arrangements, the benefit is really very low. For a simple system in sheds like this (3 rows of 16 modules in landscape on each table) the annual results (at Marseille ) are shown here for different optimizer technologies. NB: most optimizers on the market are "Buck only". Only SolarEdge uses Buck/Boost.

The irradiance incident on a tilted plane has basically 3 components: the beam, the diffuse, and a contribution of the reflexion of the sun's rays on the ground in front of (but far) from the collector plane. These 3 contributions are taken into account in the transpostion models. Now with bifacial systems, we work with the albedo of the ground just below the PV plant. This may be different from the albedo of the far terrain in front of the PV system.

Just a remark concerning your doubt about the warning "You should define at least one string per inverter or MPPT Input". This corresponds to the SubArray you are just defining: if you define N MPPT, you should obviously define at least N strings, i.e. at least one string per MPPT defined within this subarray.

PVsyst is a software for the simulation of PV systems, based on PV modules characteristics. The main parameters used for the establishment of the PV module model are the STC electrical values, i.e. mainly reflecting the efficiency of the module . This efficiency is obviously dependent on the coverage (glass) covering. But this is not dissociated from the efficiency of the PV cell itself. In other words, the glass transparency properties are included in the STC specification. There is no specific study of the glass transmission in PVsyst. Now the simulations are dependent on the incidence angle of the sun's rays. For this we define an efficiency curve, defined with respect to the efficiency at normal incidence. This curve may be measured, but this is a difficult measurement. We strongly trust that this behaviour follows the optical laws of Fresnel. See our FAQ https://forum.pvsyst.com/topic/1190-how-to-determine-the-iam-profile/?do=getNewComment

The LID is a loss which arises in the first hours of the array exposition. Therefore it is "present" in any simulation, whatever the simulated year. The aging is applied to each year, as a "yearly degradation". The loss due to the aging for the first year is the average between the beginning and the end of the year. I.e. a yearly loss parameter of -0.4% will induce an aging loss of 0.2% for the first year, 0.6% for the second year, etc. The value mentioned on the datasheets have nothing to do with the average aging degradation. These are the warranty of power for any module in the array. Please see the help " Project design > Array and system losses > Ageing, PV modules degradation "

With a PNom ratio of 1.25 you are at the limit of the overload. The global yearly contribution of your peak shaving savings is 33.6 MWh / 18’930 MWh, i.e. 0.18%, i.e. negligible. This cannot motivate the expense of a Peak shaving storage. During these months, you never attain conditions when the system output power exceeds the Power limit, so that you don’t have any peak shaving.

First of all: the one-diode model for the low-temperatures is correct down to about -10°C. With most modern modules, the one-diode model fails at very low temperatures, the voltage thermal coefficient is changing sign, meaning that the voltage is decreasing instead of increasing when the temperature diminishes. This may lead to an error in the Voc(TMin) evaluation for the max. nimber of modules. Please check with a linear model (for example by explicitly setting the muVoc value in the PV module definition (page "Additional data"), and specifying that the program should use this value in the "Project's settings"). Now the limitation of such a configuration is mostly dominated by the Power overload. When Power overload occurs, the inverter will move the operating point on the I/V curve towards higher voltages, therefore lower currents. This explains that the current limitation is almost never reached in usual configurations, it is always "preceded" by a power overload. Attaining the current limit is only possible when you define very short strings, at the limit oif the minimum voltage of the inverter. The PVsyst simulation completely manages all these situations. This is explained in the Help " Physical models used > Grid inverter > Inverter Operating Limits"