Grid-connected photovoltaic installations
4.3.2.3 Optimisation of the final system
Assuming that the owner of the shop can finance the purchase of traditional crys- talline panels, we shall look more closely at the losses and attempt to improve the forecast production. Figure 4.16 details these losses.
There are three modifiable losses. ● Module quality loss
If the supplier can guarantee the power of the modules delivered, this loss may be reduced to 0.
● Array ‘mismatch’ loss
This arises from variations in panel characteristics in terms of currents and voltages different from their maximum power values. If flash tests are available, the panels can be better paired to connect panels with identical currents in series in an
attempt to achieve a uniform Vmp voltage between the different strings. This loss
can be reasonably assumed to be reduced to less than 1%.
Table 4.5 Solar roof – three possible choices
Panels
Supplier Sulfurcell Solarfabrik Sunpower Unit
Type SCG60HV SF130/4 SPR 315 Power 60 130 315 W Module Vmp(STC) 41.5 17.72 54.7 V Module Imp(STC) 1.45 7.34 5.76 A Number in series 12 22 12 Generator Vmp(60C) 423 340 582 V Generator Voc(–10C) 684 541 838 V Parallel strings 44 20 21 Quantity 528 440 252 Generator power 31.7 57.2 79.4 kW Inverter
Supplier Solutronic Sunway Sunway
Type SolPlus 300 TG 75-ES TG82/800
Power 30.0 50.0 63.0 kW
European efficiency factor 96.0 95.8 95.5 %
Simulation
Performance ratio 83.2 79.9 79.8 %
Annual energy produced 36.5 63.2 87.7 MWh/year
● Ohmic cabling losses
These losses are easily manageable by installing adequate diameter cabling and doing away with the anti-return series diodes often installed in strings of panels. We saw in Section 3.1.4 that the panels are protected against hotspots by parallel diodes cabled every 15–20 cells. These diodes are sized to support directly the current equivalent to 1.5 or 2 times that of the panel. But if several strings are connected in parallel, when a hotspot occurs the reverse current supplied to the shaded zone can come from all the strings and exceed the panel current, which can burn out a protective diode causing the destruction of unprotected cells. To avoid this problem, the diode is sometimes installed in series with another string of panels, causing a loss of voltage. The alternative solution suggested by the manu- facturers of inverters is to install a fuse with a value higher than the maximum direct current of each string, so that in the case of a hotspot, the reverse current would blow the fuse before reaching the maximum value of the parallel protective diodes. Figure 4.17 shows a multi-string connection box being installed: each string is protected by a fuse and a needle galvanometer to show the current; it can immediately be seen if one string is supplying less current than the next one. The assembly is also monitored remotely through a modem by the supplier of the inverter, who will warn the manager of the generator if one string is supplying less power than its neighbours.
Total horizontal irradiance
Diagram of losses over the whole year
+ 10.1% – 3.5% – 5.8% – 2.0% – 3.3% – 2.2% – 1.5% – 4% –0.0% –0.0% –0.0% –0.0% 1 257 kWh/m2 63,204 kWh 65,850 kWh 76,634 kWh
IAM (incidence angle modifier) factor adjustment STC efficiency = 13.2%
1 336 kWh/m2× 433 m2 collectors
Loss due to the level of irradiance Loss due to temperature of array Loss due to module quality Array loss due to mismatch Ohmic losses in cabling
Energy from generator, virtual output at Maximum Power Point (MPP)
Inverter losses during operation (efficiency) Inverter loss, overload
Inverter loss, power threshold Inverter loss, overvoltage Inverter loss, voltage threshold Energy output from inverter Effective irradiance on collectors PV conversion efficiency at STC = 13.2% Adjustment for angle of tilt
This string connection box has components typical of this equipment (Figure 4.17).
● On the upper left (1), two variable resistances are fitted as lightning conductors
(plus and minus poles earthed).
● On the upper right (2), the main circuit includes the string galvanometers and
fuses.
● Below left (3), the main output cable box connecting to the inverter.
● To its right (4), the general DC switch controlling the input to the inverter.
● Next (5), the negative terminal strips.
● Finally on the right (6), the circuit containing the current sensors with com-
puter processing of the data for remote monitoring.
Looking again at the simulation, we will enter a loss of 0% for minimal effi- ciency and 1% for mismatch. For the ohmic losses, we would choose the fused connection box model and estimate the average lengths of the cabling.
● Each panel uses a cable of 4 mm2section and 2 m length, making 44 m for the 22
modules in the series, and the average length to the connection boxes 26 m, or 70 m for each string. The cable chosen is of the same type as used by the panels, double-
insulated halogen-free Radox cable with a 4 mm2section. The software calculates
that the resistance of 20 parallel strings corresponds to 19.3 mW.
Figure 4.17 Panel string connection box with fuses and current monitoring by modem (Solarmax system)
● Cables connecting the outputs of the two connection boxes to the inverter are
35 mm2section (STC current of 73.4 A for 10 strings in parallel), which cor-
responds to an additional 3.14 mW.
● The total ohmic losses amount to 0.8%.
We are then ready to redo the simulation, which is shown in Table 4.6. The improvement in performance is 5.1%, a significant amount when the feed-in price for electricity is high.
4.4
PV generator on a terrace roof or in open country
This type of installation is used for the biggest PV generators in the open country (Figure 4.19) or on flat roofs.
4.4.1
Installation on racks
When a large number of panels are connected, their currents and voltages must be balanced if the maximum power output is to be achieved. On the basis of the measured characteristics of the panels, panels of the same nominal current are linked in series, taking care to equalise the total voltages at nominal power of each string. The panels must have the same orientation to avoid one panel receiving less irradiance, which would then limit the current of the whole series. Figure 4.18 shows the parameters of the dimensions useful for calculating the shading of panels on racks. The objective of the calculation is to find the optimum between loss of shading and the gain in pitch for the chosen panel density: if to begin with it is
Table 4.6 Optimisation of losses of the PV generator
Losses Standard After optimisation
Performance ratio 79.9 84.0 % Annual energy produced 63.2 66.5 MWh/year Specific energy 1105 1162 kWh/kWc p p L L 1 cos(b) + sin(b)/tg(q) b q
Limit of shading S (array) S (horiz) = =
decided to tilt the modules at 30 (European optimum), the maximum installed power will be defined with shading losses of a few percent. The compromise will be a choice between maximum installed power and shading losses acceptable to the client.
We have given in Section 4.2.2 a number of rules for the installation of panels in a PV generator. Those principles remain valid here with a supplementary rule governing the distribution of panels on the surface available. In the case of systems mounted on racks on flat roofs or in the open, the exploitation of the surface available will depend on the losses allowed when the Sun is low, and casts a sha- dow from one rack onto another. To limit the effect of this shading, it is best as far as possible to connect the panels in horizontal series so that one partially shaded panel does not reduce the current of a panel completely exposed to the Sun. It is recommended to add a diode or, better, a fuse in series with each string of panels so that one string shaded at the beginning or end of the day does not affect the output of the remaining panels in full sunlight.
To optimise annual energy output in relation to the area of land or roof available, the angle that the panels are pitched at and the distance between the racks will be adjusted to take into account local climatic conditions and irradiance when the Sun is low on the horizon. By running a simulation, the benefits of various arrangements can be quickly calculated, and the best one is chosen. Obviously, if the surface area is limited, the best option will be to reduce the pitch and put up with a few percent of loss in relation to the maximum attainable without shading or limitation of surface area. The reduction of pitch has other advantages:
● the supports can be smaller and therefore cheaper;
● sensitivity to the wind is lower;
● installed power is much higher;
● efficiency under diffused light is better, a characteristic that is more suitable
for thin-film panels having better sensitivity to blue light;
● the architectural impact is diminished and therefore is more acceptable.
Table 4.7 shows an example of the exploitation of a given surface area for various places at different latitudes.
The coverage ratio is the ratio between the surface of the collectors and the surface available. The first four examples have an optimal pitch of 30%, which is
the normal reference used. For Bombay, the optimum of 20 is taken as the refer-
ence. The optimisation of the system will subsequently depend on the cost of the supports and the ease of installation. For generators in cold countries, it is usual to leave a clear space at the foot of the panels where snow can accumulate without shading the last row of panels.
The losses indicated are total for array racks of infinite length and do not take into account the geometry of the cabling. If one reckons on finite racks, the losses at the ends are smaller, and if the site lends itself to it, with the lengths of racks corresponding to a whole multiple of a string of panels, then the losses from shading can be sharply reduced by cabling the strings by horizontal rows. This is the option chosen for a recently installed 1 MW power station at Verbois near Geneva.
Figure 4.19 shows a detail of the arrays that comprise four rows of modules. Note, in particular, the supporting structure that is made up of concrete anchor points in the ground with panel supports of Douglas fir sourced from local forests.
4.4.2
Solar trajectory and shading
Figure 4.20 shows the curves of annual solar radiation and shading of PV arrays at
Bourges (central France), with panels at 20 pitch and an active surface area
of 54%. It will be seen that the shading in December is between 0% and 20% from 9 AM to 3 PM and subsequently increases at either end of the day. If each rack is
Table 4.7 Coverage ratio of generators on racks
Pitch () 30 20 10 5 Coverage ratio 45.5 54.0 69.1 81.4 Relative power 100 119 152 179 Place, latitude Hamburg (D), Losses (%) 0 1.4 4.5 7.0 53.30 Final energy (%) 100 117 145 166 Bourges (F), Losses (%) 0 1.2 4.2 6.5 47.04 Final energy (%) 100 118 146 167 Barcelona (E), Losses (%) 0 1.2 4.9 7.3
41.32 Final energy (%) 100 118 145 166 Algiers (Al), Losses (%) 0 0.7 4.0 6.2
36.34 Final energy (%) 100 118 146 168 Mumbai (India), Losses (%) 0 1.3 3.4
19.17 Final energy (%) 100 133 161
made up of four horizontal rows of panels, the loss will only affect a quarter of the system for the greater part of the winter and will disappear completely between March and September.
The pitch of the racks is also influenced by the type of energy received. At low altitudes in regions subject to fall or high cloud cover, the annual diffuse energy is greater than the direct energy, and in this case a flatter angle of pitch will result in more diffuse light being collected, and the reduction in performance compared to the optimal is slight; at high altitude on the other hand, direct radiation exceeds diffuse radiation, and when snow is present, the higher angle of pitch will allow more energy to be collected in winter and will reduce problems of snow covering the panels.
4.4.3
Trackers
A tracker is an array of panels mounted on a movable surface, which follows the trajectory of the Sun on one or two axes. With only one axis, the panels can be mounted in one plane pitched at a fixed angle on a vertical post, which will point the PV array in the direction of the Sun throughout the day. The single axis can also be in the plane of the inclined panels that will then swing from east to west fol- lowing the Sun. A dual-axis tracker, more complex mechanically, will keep the plane of the panels perpendicular to the Sun whatever its position in the sky. The orientation of the plane perpendicular to the Sun’s rays is achieved by measuring the current of four photosensors arranged on either side of a shading partition
1. 22 June 2. 22 May–23 July 3. 20 April–23 August 4. 20 March–23 September 5. 21 February–23 October 6. 19 January–22 November 7. 22 December ⫺120 ⫺90 ⫺60 ⫺30 0 30 60 90 120 0600 0700 0800 0900 1000 1100 1200 1300 1 2 3 4 5 6 7 1400 1500 1600 1700 1800 1900 Azimuth (°) Sun ’s height ( °) 90 60 75 45 30 15 Shading 20%
Shed Mutual Shading at Bourges (Lat. 47.0°N, Long. 2.2°E, alt. 161 m) Plane: tilt 20.0°, azimuth 0.0°, sheds: pitch = 5.55, width = 3.00 m, Top band = 0.00 m
Shading 40%
Shading limit, angle = 20.6°
0500
separating them (Figure 4.21). The motors or hydraulic systems governing the orientation are activated by the differences in current of each sensor: as soon as one of them is slightly shaded, the system will move to compensate for this loss and reorient precisely to the axis of the Sun.
Naturally such systems are more attractive when there is much direct sunlight, in less cloudy climates. In recent years, trackers have been mainly used in Spain and Portugal where the climate is more suitable.
When the Sun is hidden by clouds, the systems can be repositioned to the horizontal to capture the maximum of diffuse energy. But often when it is cloudy, the system follows the Sun according to a pre-programmed pattern, the pitch and orientation being defined by the hour and day of operation. The trackers are sometimes lowered to the horizontal when strong winds are expected, to reduce mechanical strain.
For the sizing of such a system, it is advisable to use software that takes account of shading: this will quickly determine optimum performance in relation to the investment budget and the area of land available. Again, the following calculations have been obtained from PVsyst software that provides for this type of calculation.
4.4.3.1 1 ha available
Let us assume that a farmer in the south of France (solar radiation statistics of Nice) has a small parcel of land of 1 ha lying fallow because the soil is too stony and poor for crops. However, the field is close to a line of electricity supply and could easily be the site of PV panels connected to the grid. We compare two solutions for the installation: a generating plant with panels on racks at a fixed pitch, and one with
trackers of around 90–100 m2of panel surface. At present, the French feed-in tariff
is limited to 1500 peak hours/year, but this limit may be removed and the tracking system would then become more attractive. To compare these two installations, we’ve chosen the same panels and inverters, and the rack mounted generator will be an assemblage of several small systems.
The plot of land is 100 m2facing due south. We assume that PV arrays are both
linked to a local inverter mounted below the panels in a watertight box. Thus, the whole surface of the plot of land is utilised. Only a transformer for connection to the grid is placed at the north of the plot behind the panels.
Sensor 1 Sensor 2
Shading partition
Table 4.8 Basic system, either fixed or with a dual-axis tracker Description Systems Gains and losses Simulation No. Power (kW) Dens ity (W/m 2 ) Inclin. (%) Shading (%) Tot. (%) PR (%) Energy kWh/kWp Rel. (%) System with 30 pitch 1 13.3 110.8 13.4 0.0 13.4 84.1 1402 100 .0 Dual-axis tracker 1 13.3 78.7 50.3 0.0 50.3 86.1 1904 135 .8
Basic system
The plot of land will be covered with multiple systems of a basic tracker. We have chosen for this simulation, polycrystalline panels manufactured by Atersa supply-
ing 222 W for a surface of 1645 900 mm. These are mounted on trackers in
landscape format 12 wide and 5 high, giving a surface of 11 m 8.25 m, including
the fixing bolts between the panels. The power of this array of 60 panels will be
13.32 kW for a surface of 91 m2. It is linked to the grid through a Danfoss TLX
12.5 K inverter, a recently developed device with a European efficiency factor of 97%.
Table 4.8 shows the results of the simulation of the system first at a fixed pitch
of 30and second on a dual-axis tracker accurately following the Sun’s trajectory.
The data presented will provide a reference when more systems are mounted close together and will be affected by shading. In detail, the data show
● in column 3, the total installed power;
● in 4, PV power by plot surface;
● in 5, 6 and 7, the optical gains and losses due to inclination and shading (here
zero for a single system);
● in 8, the performance ratio for the ohmic losses and those due to pairing of
panels;
● in 9, the annual energy density;
● in 10, the energy variation in relation to the reference surface (here fixed
at 30);
● in 11, the annual energy production.
It will be noted that the dual-axis tracker gain is excellent, reaching 35.8% at this latitude and climate (see column 10). The optical gain from the pitch of the PV
array is 13.4% for 30fixed pitch due south and 50.3% when the tracker keeps the
panels perpendicular to the Sun’s rays.
Use of space available with panels on racks
The basic arrays of 60 panels (the same as above) are this time mounted in portrait format 15 modules wide and 4 high. Seven systems can therefore be mounted on a rack with a total width of 95 m, which leaves a gap of 2.5 m free around the PV
array. Each rack thus has a power of 7 13.32 ¼ 93.24 kWc. Figure 4.22 shows
the appearance of the complete system with shade cast at 8 AM on 21 December for
a variant of six racks at 30 pitch and spaced 18 m apart.
Table 4.9 shows simulation results for four variants of use for the 1 ha field available. We have calculated successively the energy produced for six to nine racks and then sought the optimal angle producing the maximum energy for the final simulation.
It is arranged in the same way as Table 4.8, the example given without shading. We have kept for comparison with a smaller system of 560 kW (six racks of 93.24
kW) at optimum pitch of 30and producing the smallest shading losses. It will be
noted that the increased density of racks does not produce too many losses, the
3.1% less efficient than our reference example, while the installed power has increased by 50%.
Optimum performance is achieved when the racks are well spaced and produce little shading, but the financial optimum is probably for a high density and better use of the available space, as the costs of preparing the ground and connecting to the grid as well as engineering and other infrastructure expenses will not change very much according to the number of racks.
Use of space available with trackers
The problems of shading are much more difficult to resolve when the trackers increase their tilt towards the horizon when the Sun is low. Shading appears at the bottom of the arrays or on the sides, and the cabling of the strings should be opti- mised according to the simulation, which enables the impact of this shading to be seen directly. The simulation will also be valuable for the financial optimisation of the investment.
Figure 4.23 shows the example of the 36-tracker variant of the generator and the impact of shading at 8 AM on 21 December.
Table 4.10 shows the simulation results systems with 25, 30 and 36 trackers on