PURPOSE : 800 kWp PV Power Plants for Direct Injection in Light Train Low Voltage D.C. Networks
ABSTRACT : The purpose of this project is to manufacture and install such PV power plants in Germany, 250 kWp at Hannover and in Switzerland, 154 kWp at Geneva and about 100 kWp at Lausanne.
New modules with specifications classII 1500V are certificated by some europeen manufacturer for the Heliotram project. Application from industrial safety for photovoltaic high system fiability. Use of ultra-fast bidirectional swich with special magnetic coil, HF Filter and lighting protectection. Fly wheel energy storage systems are installed in the Hanover LRT system to intermediately store recovered braking energy until an approaching train can use it. They can also be used for the intermediate storage of solar power generated by the PV plant which is not being utilised. All the data of this plants are equipeted with a special monitoring.
KEYWORDS : 650VDC grid connected, new modules homologation, industrial plant, building integration, Magnetic Dynamic Accumulators. SE 146 / 96 DE-FR-CH
Project web site: www.sunwatt.ch
Figure 1a : Hanover 250KWp
Figure 1b : Geneva 154 KWp
Public transportation low voltage D.C. networks are particularly well suited for direct connection of PV systems because of their electrical characteristics, since the load of the network coincides with the PV production. Very simple and therefore reliable PV power stations can be built to inject the solar production into such networks, without power conditioning.
In the framework of the Thermie 96 programme, Sunwatt presented and developed the project called Heliotram, which is aimed to the connected PV large Plants for injection in light train low voltage D.C. networks. Since 1992 a small plant of 2kwp are connected to theD.C. networks of the railway and give us some data and experience to have a high system fiability.
The purpose of this project is to manufacture and install such PV power plants in Germany (250 kWp at Hannover) and in Switzerland (154 kWp at Geneva and about 100 kWp at Lausanne). This implies:
- Direct connection of series of PV modules to 600-750V DC grid, i.e. peak voltages up to 1'000V at open circuit by cold weather.
- Incite the manufacturers to certify their PV modules for a high voltage usage. The PV modules in use will be continuously exposed to high voltages and thus require a good insulation between frame and cells.
- Testing high speed circuit breakers and power contactors, for DC high voltages and a large intensity scale (0.5A to 250 A bidirectional, under 1'500V).
Swiss industry is in a pole position in this area. For the PV modules we incite several manufacturer to certify their modules for a high voltage usage, solarmodule class II. The same high voltage certification holds also for the quick connections and other materials.
The system of TPG of 154 kWp, at Geneva, has started operating in July 1999. First results on the system behaviour and performances are presented.
2. Installations Description
2.1 Ustra in Hannover :
The PV field is integrated on the new-built Leinhausen Depot of the Üstra company of public transports at Hannover. The Hannover's plant has benefited of development realised in Geneva in the choice of components, security systems and monitoring. So the concept of Hannover is the same than this of Geneva. (fig.5)
2.1.1 Photovoltaic modules :
230 kWp (1615 modules of monocrystalline silicium) are integrated on the roof of the building. This PV field is composed of 8 long sheds of about 80 and 100 meters long, and 3 little ones on the central part of the building. These fields are oriented at 35° from south direction (toward West). Five of the long sheds are coupled with light dwells intended for hall ligthning on their nordern face. The PV collectors of these sheds are tilted by 20°. The other sheds are placed lower on the roof, and tilted at 10° only.
A complementary field of about 19 kWp is placed on the south facade. This part was primarily foreseen with amorphus modules. But the evaluation of the offers revealed that there are no suitable technologies on the market, meeting the demands for :
- Voltage stability,
- Building integration and
- Efficiency guarantees over the service life.
Therefore this part was also equipped with crystalline-Si cells modules.
The contract considers the application of the following modules as creating the photovoltaic generator (table 1) 2.1.2 Combination with Magnetic Dynamic Accumulators
Magnetic dynamic accumulators will be installed in the Hannover LRT system to intermediately store recovered braking energy until an approaching train can use it. They can also be used for the intermediate storage of solar power generated by the PV plant which is not being utilised.
A test installation at another location on the LRT network (Fasanenkrug terminus) confirmed the proper functioning of the system. The operating behaviour of the magnetic-dynamic accumulator used in the test will however require some optimisation of the technical layout.
Magnetic-dynamic accumulators are also planned for the Leinhausen Depot. It is specified with the power characteristics 3,3 MW / 9,2 kWh at a charge / discharge cycle of 30 seconds up to several minutes The construction provision required has already been undertaken. According to üstra's installation plans for their magnetic-dynamic accumulators, they will not be available in Leinhausen at the same time of commissioning the PV plant. This has no direct relevance in this context because they are not part of the project.
|Module Type ||Manufacturer ||Cell/Module Type ||Module Power (Wp) ||Number of Modules ||Generator-Power (kWp)
|Insulation Glass Modules
||Siemens SM 103
||MST 43 L
Table 1: Hannover photovoltaic modules
2.2 TPG - Geneva :
The Geneva plant is also located on the roof of the Public transport Depot. Nevertheless the metallic roof was already built about five years ago, and the PV field had to be disposed in sheds on dedicated supports, with strong requirements about the weight and the waterproofness. Therefore, an unusual flat tilt of 5° (minimum necessary to the water flow and dirt removal) was chosen, leading to about 8% irradiation decrease by respect to optimal tilt of 30°. Nevertheless this allowed to drastically reduce the weight and costs of the supports.
The field is made up of 1'400 mono-crystalline modules of 110 Wp each, i.e. 70 strings of 20 modules in series. In order to match the nominal line voltage of about 640V, the optimal number of modules in series was found using the PVSYST software, which is able to simulate such a DC-grid installation.
2.3 TCL - Lausanne :
Beyond the Geneva part of the Swiss contribution, three little installations have been implemented in Lausanne, these are :
2.3.1 Perrelet plant, 19 kWp installed as a sun-shield on the top of the city Bus Depot. These are 396 modules Photowatt PWX500T, tilted about 20° in south direction. For a nominal line voltage of 660VDC (+30%/-20%), 44 modules are connected in 9 series strings.
2.3.2 Riponne plant, 10.7 kWp of the same module type, as sun-shields for shop windows. Orientation is comparable to the precedent, but standing against a facade. The number of modules in series is here 45 per string.
2.3.3 Rhodanie plant, 14.4 kWp of "Powerwall laminates" 240 Wp modules disposed on a south façade.
These three plants will start working at beginning of march, 2000. Security concept of these installations is undertaken by Lausanne.
3. Early performance: measurements for Geneva :
For such a high voltage plant, the module specifications have to undertake an insulation voltage of 1'300VDC. Modules are disposed in 7 sheds of 200 modules each, with a very low tilt (5°). Due to the low tilt, the modules are frameless to avoid water, dirt and mosses accumulations. Only one shed has been equipped with framed modules in order to compare the results over a long term.
Connexions of each shed are collected in a coupling box, including safety diodes, para-lightning protections, swiches, and current measurements for each string. Power is driven to the main power cupboard through about 200 m cables. There is no MPT nor other power conditioning system: the array is directly coupled to the trolleybus DC-line, through an ultra-fast swich (URL) and a HF filter.
A monitoring system records all relevant parameters, including the 60 individual string currents, with a very short time step. Data can then be collected in hourly or sub-hourly values for system operating analysis.
3.1 System operation and monitoring :
The Heliotram power plant operation started in July 1999. The operating conditions were satisfactory during the 6 first months up to now. No serious defect was encountered.
The main following tests were conducted during the starting-up :
- Current leakage during an insulation voltage test at 5 kV (the PV modules of Class II are required to be certified for an operating voltage up to 1300V, and the module "fast" connectors has been certified for operation up to 1500 VDC).
- Test the high speed circuit breakers and power DC contactors for high voltage and a large scale of intensities (0.5A to 250 A bidirectional, under 1000VDC).
Monitoring is performed by a PC computer; data are collected all over the array through decentralised data acquisition modules, interconnected by a RS485 bus. The monitoring system records the following data :
- Array voltage
- Grid voltage at injection
- Plus and minus branch global currents
- Individual currents of the 60 module strings
- Fuse and lightning protections states in the connexion boxes, and other general system protections,
- Disconnector states
- Box temperatures, module and ambient temperature, etc.
In order to ensure an early detection of any system misfunctioning, a very fast acquisition rate was chosen, resulting in a great amount of data. For this report, we have 3 periods of detailed data.
3.2 System Performance Data :
Putting together the meteo data from the GAP, and the TPG detailed and hourly data, we obtained 60 days of clean data which have been carefully analysed.
These data are summarised on the table 2 and fig 2. They are presented using the reduced "universal" normalised variables proposed by JRC/Ispra (ref ), which allow to show properties of the system performances itself, independently of the geographical situation, the plane orientation or the system size.
3.3 Input-output diagram
The system's instantaneous behaviour can also be visualised as the Input/Output diagram(fig.3), which displays the System Output energy as a function of the Incident Irradiation in the collector plane.
The linear adjustment indicates the average Power Output, which will be 112 kW for an irradiation of 1 kW/m2.
Dividing by the Module rough area (including frames) indicates the global system efficiency , that is 9.4%.
We can see on the diagram that the dependence is not quite linear: the system production decrease for higher powers, essentially due to temperature effect. This behaviour will be analysed in the next section.
4. Data analysis
4.1 Wiring ohmic loss
Wiring cables have been largely sized in order to minimise the ohmic losses. The resulting array resistance is calculated as about 30 mOhms Moreover, a HF filter including two coils in series had to be inserted in the coupling connections with the grid. (20 mOhm).The only way to estimate the real loss is to perform simulations over one year with different values. This shows that the overall ohmic loss is 1.5% for the cabling and the coils together.
4.2 Mismatch effects
The mismatch loss is strongly dependent on the operating voltage. The mismatch loss has been estimated at 4%.
4.3 Number of modules in series
The number of modules in series was optimised using PVSYST during the design, and was chosen as 20 modules. This optimum corresponds to the maximal yearly available production (or maximum system yearly efficiency).
With this optimum, the winter operating point will be below the MPP voltage, and during high irradiation in summer time it will pass over this point. The exact optimal balance of course requires the detailed simulations performed by PVSYST.
But by respect to design time, two voltage-sensitive parameters had to be revised according to measured data :
- The line voltage had been measured over a long time during an early PV experience on the same site (GAP, Univ. of Geneva), several years ago, and was found to be 630V on an average. No systematic measurements are available at the TPG, but they claim that they didn't modify the line. Nevertheless, the actual measured value is around 650V.
- For simulating the module temperature, we also used the usual thermal loss factor of 29 kW/m2K. But as discussed in section 4.4, on one hand the roof temperature increases with irradiation, and on the other hand the module ventilation is not so efficient as for tilted modules. Combining theses effects while using with true "meteorological" Tamb, the resulting equivalent k-value should be about 19 W/m2K.
Fig.4 shows that with this new parameters, the optimal is now 21 modules in series, which should give a yield of 1002 kWh/kWp/year.
Nevertheless the optimum curve is rather flat. The present setup of 20 modules in series should yield 989 kWh/kWp/yaer, i.e. a production loss of about 1.3%
by respect to the optimum.
With Unom = 650V and k = 19 W/m2K. These simulations also quantifies the loss due to operating fixed voltage by respect to the MPP operation. This is limited to about 5000 kWh/year, i.e. about 3.3% of the production.
The project HELIOTRAM - TPG, comprising 154 kWp of PV-modules directly coupled to the DC grid of the public transports of Geneva, has started production at the beginning of July 1999. The plant had been designed and optimised using the PVSYST software simulations, and a production of 1038 kWh/kWp/year had been foreseen (i.e. a total production of 160'000 kWh/year).
Up to now , the plant operated satisfactorily. About 60 days of detailed monitoring data have been analysed.
During design, two essential parameters were slightly misestimated. On the one hand the grid operating voltage has been raised from 630 to 650 V since our old measurements. On the other hand the thermal factors didn't take into account that cooling by ventilation is less efficient for horizontal modules than for tilted ones. Moreover, due to an asphalt covering, the overall roof temperature strongly raises with insolation.
These observations will lead to a yearly loss of about 5% by respect to the previsions, i.e. about 990 kWh/Kwp/year. Wiring ohmic losses are limited to 1.5%. Other losses of less than 3%, could be attributed to the module quality or mismatch by respect to the manufacturer specifications.
However, the real measured yield of the plant stays within some few percents of the foreseen production. The result of about 1000 kWh/kwp/year is far over the Swiss average yield of PV plants, and is a remarkable result for such a module layout, almost horizontal, which is one of the main special features of this plant.
 Guidelines for the Assessment of Photovoltaic Plants :
Document A Photovoltaic System Monitoring
Document B Anlysis and Presentation of Monitoring Data.
CEC/JRC : Commission of the European Communities, Joint Research Centre, Ispra (Italy),June 1993.
 Pierre Ineichen, GAP, University od Geneva, Private communication.