Why big data is a big deal in smart PV

The process of collecting and analysing large volumes of data from PV power plants is at the heart of the smart PV concept. It offers plant operators previously unimagined capabilities in running assets optimally and above all profitably.

Behind the need to replace fossil fuels with renewables and replace high electricity consumption and wastage with more energy efficient systems, there will need to be new kinds of grids.

Or rather, grids will have to adapt. Handling distributed generation resources like solar, wind, electric vehicles (EVs), demand response and energy storage, co-ordinating them from high voltage transmission level down to PV plant monitoring to the smart meters that are being increasingly rolled out is going to be a tough balancing act.

Solar advocates and industry groups rightfully point to reluctance from some utilities to embrace renewables, sensing an existential threat to business models and revenues. However, it is also undeniably true that integrating variable renewable energy, coping with new technologies like rapid chargers for EVs and keeping the distributed network of the future stable presents a challenge.

“It seems that every day we read new stories of declining electricity demand due to the rapid expansion of both energy efficiency and distributed generation. Over the past decade, consumers have increasingly adopted distributed energy resources (DERs), which has had significant impacts on the bulk power grid and distribution operations,” says Omar Saadeh, grid and distributed resources analyst at GTM Research.

“Amidst all this substantial ongoing change, utilities are mandated with maintaining the delivery of reliable electricity – and they’re rightfully concerned. The variability of distributed and renewable generation produces significant operational challenges, such as two-way power flows, balancing discrepancies, and so on – creating a higher demand for rapidly deployable grid flexibility,” Saadeh says.

The key to a smarter grid will rest therefore on how well the grid can respond and how not only its operators, but also the constituent branches of the distributed network, deal with information and react.

‘Big data’ is a relatively loose term that either means more than a terabyte – 1,000GB – of data, or can be taken to mean data too rich and vast to be processed by conventional techniques. In the instance of the grid and distributed energy resources, it is apt for the masses of inputs and variables that need to be taken into account.

Components of a distributed network, while not necessarily tied to each other with power cables, will be tied together by IT. To drop in another currently popular phrase, a network that can react predictively and in real-time to technical considerations and electricity consumption and production patterns will equate to – as one company interviewed called it – an “energy Internet of Things”.

Big data and smart O&M

The concept of big data lies at the heart of many of the capabilities offered by smart PV systems. It is particularly critical to the efficient operations and maintenance functions that can be achieved through the cloud-enabled remote monitoring and management of distributed power plants. It allows discrete analysis of components to go on continuously and for most O&M functions to be carried out with either no or very few personnel on site.

In the smart PV solution, the inverter performs high-accuracy sample measurement of the voltage, current and other working parameters of each string in the array, with the measurement accuracy reaching 0.5%. Utilising the communication system of the power plant, engineers can remotely check the working status and parameters of each string at the background at any time, implementing remote maintenance and smart O&M.

If an inverter or a string is abnormal, the smart monitoring system proactively reports alarms, and locates faults in a quick and accurate manner. In the entire process, the operations are safe and power interruption is not required, which minimises the preventive maintenance and O&M costs without affecting the electricity generation.

By comparison, in power plants where a central solution is deployed, the combiner boxes and inverters are most likely to be supplied by different manufacturers. The quality of these devices is hard to control, and the communication between them is difficult to manage; communication between combiner boxes may fail due to inaccurate or even incorrect signals. Therefore, it is difficult to know precisely the working status of each string. Even if exceptions can be detected by other means, it is difficult to quickly and accurately locate and resolve problems.

The current method for checking the working status of each string in the PV area is as follows: locate each DC combiner box in the area, open the combiner box, and use a hand-held clamp ammeter to measure the working current of each string. In some power plants, however, as the DC cables inside the DC combiner box are densely distributed, the clamp ammeter cannot be inserted to measure the working current. In this case, the O&M engineers have to disconnect the switch of the DC combiner box and the fuse of the corresponding string, and then measure the voltage and check the fuse status of one string after another. Therefore, field O&M is difficult and slow. This brings a heavy workload to O&M engineers, increases skill requirements and may even endanger the personal safety of O&M engineers.

Fault detection

After a power plant is put into commercial use for a period of time, faults caused by various factors will gradually occur.

In most centralised power plants, it is hard to detect string faults, and the loss of electric energy yield cannot be compensated. Based on the high-precision string-level detection of the smart PV controller, the system can detect faults in a timely manner. By analysing databases, the system can accurately locate a specific faulty device and propose handling suggestions based on preset measures and O&M experience.

In power plants where the central PV solution is deployed, if modules, wiring during construction and natural factors are not considered, faults of DC combiner boxes and inverters are the major factors that cause loss of electricity.

Faults of DC combiner boxes account for a large proportion among all the faults in a PV power plant. Assume that a string contains 20 250Wp modules and the total power is 5kW. In a 1MW solar sub-array, the electricity lost due to the faults of fuses of a string accounts for about 0.5% of the total electricity generated by the entire sub-array. If a combiner (with 16 inputs and one output and a total capacity of 80kW) is faulty, all the strings connected to the combiner box cannot generate power properly, which will result in the loss of about 8% of the total electricity generated by the entire sub-array. Due to the low communication reliability of the combiner box, it is difficult for O&M engineers to detect and rectify faults as soon as they occur. Most faults are often detected during the preventive maintenance or when the accumulated impact is large. At this time, however, a great loss of the generated electricity is already caused by these faults.

If power generation is affected due to the fault of an inverter, about 50% of the total electricity generated by the entire sub-array will be lost. Central inverters must be maintained by professional personnel. The accessories of inverters are large and heavy. It takes a long time to detect, locate and eventually rectify a fault. Assume that a 500kW inverter converts electricity for four hours every day, and it takes three to seven days from occurrence of a fault to rectification of the fault. Therefore, when this inverter is faulty, the electricity lost due to the fault of the inverter is 500kW x 4h/d x 7 d = 14,000kWh. The economic losses from such a scenario quickly add up.

In power plants where the smart PV solution is deployed, if modules, wiring during construction and natural factors are not considered, only faults of DC combiner boxes will cause loss of electricity because the PV power system does not require the DC combiner box or fuse. Therefore, the system reliability is greatly improved. The smart PV controller is small and light. Generally, spare parts are available on site, and the faulty parts can be replaced on the very day that a fault occurs. Assume that each string contains twenty 250Wp modules connected in serial and the power of a string is 5kW. When an inverter is faulty, at most six strings are affected. Even if all the six strings generate electricity in full load, and electricity is generated for four hours every day, the loss of electricity caused by a fault of the inverter is about 5kW x 6 x 4h/d x 1 d = 120kWh. The economic losses are much reduced.

Fault rectification

The fault rectification difficulty is different due to different characteristics of the two solutions. The central inverter and many modules inside the inverter are large and heavy. Some modules even weigh dozens or hundreds of kilograms. This brings great inconvenience and difficulty to maintenance and repairing.

The smart PV solution requires no DC combiner box, and the possibility that an AC combiner box becomes faulty is almost zero. Some power plants even do not use combiner boxes, and directly connect the AC output of the smart PV controller to the bus on the low voltage side of the box-type substation. Therefore, if the smart PV solution is deployed, equipment faults are mainly the faults of the smart PV controller itself.

Compared with the huge central inverter, the smart PV controller is small and light. Two workers can complete unpacking, assembly, and wiring of the smart PV controller without presence of professional personnel. When a fault occurs, maintenance engineers can immediately replace the faulty smart PV controller with the spare parts based on the accurate alarm information, so that the power plant can be recovered within a short time and the loss of electricity can be minimised.

This article draws on material first published in Volume 05 of PV Tech Power. Reproduced with permission from the publisher.