Solar Energy Radiation Forecasting Methods in the Current Scenario Jai Singh Arya*, Omveer Singh**, and S. K. Aggarwal *** * *** Electrical Engineering Deptt., Maharishi Markandeshwar University, Ambala, Haryana, India Email: jaisingharya26@gmail.com ** IEEE Member, Electrical Engineering Deptt., Maharishi Markandeshwar University, Ambala, Haryana, India Abstract: In this article, a detailed analysis has been made for forecasting methods in the solar radiation world. Various types of solar radiation forecasting methods are shown in the article. These forecasting methods are written briefly in this article. Authors reported related merits and de-merits of these techniques in the literature review. These activities indicate that the how the solar radiation forecasting techniques importance to the green world growth. Article shows that some forecasting methods can be used alternatively or in combined way to achieve the accurate solar radiation forecasting. Keywords: solar radiation, photovoltaic, global solar irradiation, solar module, forecasting methods. Introduction Solar forecasting and Meteorology is an active field of research in solar energy forecasting for interfacing with the renewable energy. In solar energy generation and integration, the main difficulty arise due to the natural fact that solar irradiance is inherently dependent on geographical location of site, time of the year, climate in the region and level of irradiance. The fluctuating nature of energy output from these Photovoltaic (PV) sources requires reliable forecasted information for reliable integration in the energy mix of modern grid system. Solar resources are intermittent and sensitive to weather and climatic conditions. In power generation among most of the renewable energies which are widely used, the PV is one of the fastest green power alternatives. The renewable energy resources are replacing the conventional generation methods because CO 2 emissions from residential sector are increasing rapidly due to proliferation of all electric houses [1]. The behavior of solar radiation is complex, either periodic or randomly variable. The accuracy of PV power prediction is done by forecasting [2, 3]. Solar irradiance impose a significant challenge because of high variability and uncertainty, while at the same time solar forecasting and generation is subject to environmental factors which is going to be a urgency of society [4]. Due to increased penetration of solar energy into power pool, the solar forecasting plays a critical role in modern power system. Till date a variety of methods to forecast global solar irradiation (GSR) and photovoltaic output using meteorological data have been proposed. Forecasting methods use complicated processes to achieve the high forecasting accuracy. To enhance the electrical power generation and to optimize its use there is need of forecasting the solar radiation with accuracy and consistency. Current Status of Energy in the World Investigators have been observed that in nature solar energy is in abundance. It has been estimated that 174,000 terawatts of solar energy strikes the earth surface continuously which is 10,000 times more than the total energy requirement of the World [5, 6]. This Fig. 1depicts that the total energy consumption annually in the world is very small as compared to energy available annually on the surface of earth by insolation. Estimated availability of various sources of energy The coarse view the different type of energies available in the world has been shown in Fig. 1. It is noted from the estimate that there is a uncertainty of the amount left behind of various resources of energy [6]. Most of these resources are diminishing day by day as the consumption of energy is also increasing due to technological advancement. Besides this the most of the conventional resources of energy pollute the environment and creating the health hazards globally. Loss of irradiance efficiency with the varying direction of solar irradiance It depicted from the Fig. 2 shown below at optimum position of Solar Panel, the efficiency loss is 0% when Solar Panel is completely facing the Sun. When the position of solar irradiance changes from 0º to 90 o, the irradiance loss increases from
146 Advances in Engineering and Technology Annual energy consumption Gas reserve Oil reserve Nuclear reserve Coal reserve Annual solar insolation Fig. 1 Overview of potential of solar energy in the world 0% to 20% [7]. It has been analyzed that average efficiency drop of a solar panel mounted away from the optimum position is about 1.1% for every 5º degrees. To cope up with the increase in demand of electrical energy the newer technologies are being developed and also the conventional fuels used for electricity generation. Newer technologies are being developed in the whole world as well as in India also to produce the electricity from other available renewable energy sources to curb the emission of gases which are causing global warming. The utilization of the renewable energy sources has may make a sea change in power sector of India. Fig. 2 Loss of irradiance efficiency with the varying direction of solar irradiance In future, all sources of renewable energies will be harnessed and integrated with the Grid. Most of these are very effective and cheap. But this all is possible with effective forecasting of solar energy. Moreover there are problems and hindrances due to public litigations against land acquisition and relocation of land for conventional power plants. Similarly Thermal Power Plants also have their limitations like non availability of good quality coal, environmental pollution, and import of coal, its transportation cost and huge cost on bulk transmission system to transmit the electrical power from point of generation to load centers. This can be tackled by installing the solar Power plant after proper forecasting the region for availability sun rays. Nuclear power generation in India, although efficient and relatively new, has some constraints like high investment for setting up the plant, unavailability of ready-to-use nuclear fuel in the country and safety concerns due to emission of dangerous radiation. Due to these shortcomings and constraints of the conventional methods of energy generation, the
Solar Energy Radiation Forecasting Methods in the Current Scenario 147 renewable energies are to play an important role in managing the energy demand of society. To manage power requirement and the increase in demand, the use of renewable energy resources should be integrated with the grid. In future High flux density and large current techniques will increase the generating capacity of the machines. To ease the energy constraints various high power electricity storage technologies such as capacitors, flywheels, with renewable resources in order to achieve clean and green power of high quality, flow batteries to store the solar energy, which will provide fast response voltage and better power quality management. Due to high insolation in Indian region the use of solar energy has increased. During the last five years the solar energy generation has increased from 161 MW to 3100 MW in 2015. In future, it will be mandatory to install and use of solar powered equipment on all the govt. and private buildings. Electricity Act 2003 has two main objectives; one is electricity supply for all areas and second promotion of renewable technologies for clean and green environment. The monopoly of the supply companies will also be affected by this act. Classification of Forecasting Methods a. Stochastic Learning techniques b. Artificial Neural Network c. Numerical Weather Prediction Method d. Satellite Image techniques e. Ground based image techniques Stochastic Learning Techniques of Solar Forecasting In these techniques, the pattern of DATA is identified within one variable (e.g. Auto regression) and between variables or even images. The assumption of future irradiance can be predicated by studying the historical data or patterns. The simplest stochastic model learning technique is the persistence forecast which based upon the recent output of PV power plant or radiometer output. The data obtained is extrapolated to account for the changing sun angles. Stochastic Modeling is univarate and multivariate forecasting of hourly solar radiation with artificial intelligence techniques [8]. Artificial Neural Network The development of Artificial Neural Network (ANN) began with the name Artificial Intelligence in 1950. It is basically combination of pattern recognition, deductive reasons and numerical computations to simulate. In ANN techniques, the climatologically variables taken as input and to predict the monthly values of global horizontal irradiance over a year. Numerical weather prediction method The current weather condition are considered as the input describing as the processes occurring in the atmosphere to predict the weather for a certain future period in this method. The Numerical Weather Prediction (NWP) is applicable for a fore casting of solar irradiance. The Autoregressive Moving Average (ARMA) and Autoregressive Integrated Moving Average (ARIMA) are very useful techniques for the calculation, analysis and optimization of the non stationary processes. Due to this, these techniques are also used to forecast the solar radiations also. Satellite Images Techniques It is similar to total sky method. In this method, clouds reflect light into the satellite leading to detection and the amount of light transmitted through cloud is calculated accurately. The lower value of spatial and temporal resolution causes satellite forecasting to be less accuracy than sky imagery on intra hour time scale. Ground based image techniques Ground observations using a total sky imager (TSI) present a clear view of cloud shadows for forecasting. Better method for spatial and temporal resolution [8]. Literature Survey Few years before, most of the PV systems installed in the world utilized batteries for energy storage and were not connected to an electric grid. In the last decade, the connection of PV system has become dominant way that the world uses this technology. However off grid PV system is also a solution to power needs in the areas where nearby grid access is not possible. The sunlight is the most abundant energy resource for human beings and out of it solar photovoltaic is a green energy which is playing a critical in ensuring the power security [9]. It is by nature that solar forecasting is playing an excellent role in supplementing the increased power from PV system. Solar power use increased at a record pace in United State of America (USA). Forecasting of solar energy consumption in USA has been discussed. Knowledge-Base library concepts were used by two formats. In first concept one minute intervals,
148 Advances in Engineering and Technology the data was collected for the entire day. Similarly images are sampled at ten minutes intervals during the day [6-9]. The share of global power production from solar power generation is increasing day by day and the same power is integrated with existing power grid. Solar forecasting method has been developed and described at technical university of Catania in Italy. The solar radiation forecasting has been taken as most significant aspect in meteorology and solar power conversion. These various methods predict the output power of solar plant and help in controlling the power generation by other conventional methods. These methods forecast two day ahead, one day ahead and three hours ahead respectively based upon the weather information. This describes that the forecast information is necessary for the management of electricity grid. The accuracy of PV power prediction is investigated with the help of case study. Major methods of forecasting have been described to predict the output power of photovoltaic system in power generation. In 2012, it was expressed that radiation prediction is a challenge for researchers as this prediction is used to estimate the power to be generated by the other commercial power generation resources. Solar radiation prediction is a big challenge to estimate the power developed in PV module. This paper analyzed with the problem of solar radiations based upon the observed meteorological data [10]. The forecasting data may be used to reduce detrimental impacts on the distribution system due to voltage fluctuations caused by the increased PV penetrations and due to high variability and uncertainty of solar irradiance. Solar generation variability affects the long term planning of the system [11]. It was stated that solar power forecasting is a key component in planning of solar PV power plant. Soft computing techniques can predict the solar generation better than conventional methods [12]. In Agra, solar power forecasting modeling was presented. It was described that artificial neural network and generalized neural network are used as powerful tool for renewable energy forecasting and prediction the solar power generation. It was also discussed that solar is the best alternative energy source as compared to other conventional sources. In the University of Texas, a method of an adaptive rules generation for solar forecasting to the full scope of solar forecasting techniques. Both present and future prospectuses have been discussed [12, 13]. In USA, the Sacramento Municipal Utility District has been working with several commercial forecasters to evaluate the Performance of their forecasts in predicting PV output and generation control. Various irradiance monitoring system were installed in the solar network to check the PV output of about 100 MW solar generation power plant. To make the forecasting useful and accurate new tools of forecasting like distributed generation forecasting were used. The analysis also done both ways the irradiance forecast and the power output forecast by which overall accuracy of the forecast evaluated. With these analyses; it become clear that the inaccuracy in forecasting increases during the high variability of cloud cover and irradiance. Although the level of accuracy was enough sufficient but is not up to mark viewing the expectations of future. Many improvements are to implemented to make the forecasting more advance in future [14]. The high temporal variability of solar power is a real issue to achieve a balanced production and consumption in a power system in IEEE magazine. Solar power forecasting is then necessary to better exploit the variability and uncertainty for smooth running the energy mix system. In this method, forecasting is done based upon the phase correlation algorithm by which cloud maps are derived by Meteosat system. PV power production highly relies on the irradiance reaching on solar panels and varies with weather condition and cloud thickness. Accurate prediction through forecasting helps the plant managers and grid operators to manage their system effectively. By this energy mix in the grid will be eased and optimized. Study of cloud cover from satellite images is superior to the other conventional methods [15]. It was found that most practical way to predict this renewable source of energy is to use meteorological data, but its exploitation is hampered by its nature of intermittency and unreliable availability. The other reason is that cost incurred in this technology is falling day by day due to research and technological advancements. In power engineering, Supply-demand balancing is always required, so a variety of methods to forecast the PV output and solar radiations are used to achieve high accuracy in forecasting. It was discussed that to improve the real time performance and reducing the negative impacts of PV power generation an accurate forecasting is required. For that it requires very accurate forecasting solar irradiance. Various methods have been proposed to improve the accuracy of short term PV power generation in solar power system. It is also concluded that during the recent years; the solar PV generation in power system has been seen as dramatic increase in energy mix. This increasing penetration of PV electrical power has posed a challenge to the reliable operation on the power management due to variable in nature. So the accurate prediction of solar PV power generation is an area of modern research and development to get the quality power with high reliability. Conclusion In this article, basically major methods of solar radiation were discussed and compared. The most physical approach is numerical weather prediction. Hybrid methods which incorporate two or more methods seem to be most promising for high fidelity irradiance forecasting in future. It is clear that solar variable generation forecasting is a major challenge since the last many years. But now it is the key component in the integration of large solar power integration, so lot of improvements in forecasting are required to provide substantial economic and reliability benefits to the system. As solar irradiance is unpredictable and irregular component in nature, newer investigations may carried out with large data collection from a solar station. Through improved forecasting the predictability is to be improved for enhanced reliability in
Solar Energy Radiation Forecasting Methods in the Current Scenario 149 energy mix system. It is aimed to capture the various spatial-temporal variations of the solar irradiance to forecast it minutely, efficiently for the better results as compared to prior results in future. Effective forecasting will also help in ensuring the electrical energy availability and preserving. References [1] K. Uchida, T. Senjyu, N. Urasaki and A. Yona, Installation Effect by Solar Heater System using Solar Radiation Forecasting, IEEE Transactions on Transmission & Distribution Asia, 2009. [2] S. K. Aggarwal, L. M. Saini and A. Kumar, Electricity price forecasting in deregulated markets: A review and evaluation, Journal on Electrical Power and Energy Systems, vol. 31, pp. 13-22, 2009. [3] E. Lorenz, J. Hurka, D. Heinemann and H. G. Beyer, Irradiance Forecasting for the Power Prediction of Grid-Connected Photovoltaic Systems, IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, vol. 2, no. 1, March 2009. [4] F. Nomiyama, J. Asai, T. Murakami and J. Murata, A Study on Global Solar Radiation Forecasting Models Meteorological Data and their Application to Wide Area Forecast, IEEE International Conference on Power System Technology, Auckland, pp. 4673-2868, 2012. [5] New Report on Global Solar Industry, 2015. Online: www.solarserver.com [6] P. Bacher, Short-term Solar Power Forecasting, Thesis, Kongens Lyngby, 2013. [7] M. Boxwell, Solar Electricity Handbook, Green-stream Publishing, Hargrave Close, United Kingdom, 2015. [8] Inman and R. Headsen, Solar forecasting review, University of California, San Diego, 2012. [9] T. Lamnguyen, Y. F. Huang, M. H. Shu and B. M. Hsu, Forecasting Model for the Solar Photovoltaics Consumptions in United States of America, IEEE Conference on Power and Energy, China, pp. 284-288, 12-14 December 2012. [10] G. Capizzi, C. Napoli, and F. Bonanno, Innovative Second-Generation Wavelets Construction with Recurrent Neural Networks for Solar Radiation Forecasting, IEEE Transactions on Neural Networks and Learning Systems, vol. 23, no. 11, pp. 1805-1815, November 2012. [11] K. Stefferud, J. Kleissl and J. Schoene, Solar Forecasting and Variability Analyses using Sky Camera Cloud Detection & Motion Vectors, IEEE Power and Energy Society Meeting, San Deigo, CA, pp. 1-6, 2012. [12] V. P. Singh, K. Vaibhav and D.K. Chaturvedi, Solar Power Forecasting Modeling Using Soft Computing Approach, IEEE International Conference on Engineering, Ahemadabad, India, pp. 1-5, 6-8 December, 2012. [13] J. Nummikoski, Y. S. Manjili, R. Vega and H. Krishnaswami, Adaptive Rule Generation for Solar Forecasting: Interfacing with A Knowledge-base Library, IEEE 39 th Conference on Photovoltaic Specialists, Tampa, FL, pp. 980-984,2013. [14] O. Bartholomy, et al, Benchmarking Solar Power and Irradiance Forecasting Accuracy at Sacramento Municipal Utility District, IEEE 40 th Conference on Photovoltaic Specialists, Denver, CO, pp. 63-68, 2014. [15] S. Cros, O. Liandrat, N. Sebastien and N. Schmutz, Extracting Cloud Motion Vectors from Satellite Images For Solar Power Forecasting, IEEE International Symposium on Geoscience and Remote Sensing, pp. 4123-4126, Quebec City, QC, 2014.