Forecasting of daily natural gas consumption on regional basis in Turkey using various computational methods
Highlights
► We modeled residential natural gas consumption regional basis various computational methods. ► We used both natural gas consumption and meteorological data including ambient temperature, etc. ► We developed a SARIMAX and two ANN models forecasting daily gas consumption in a city in Turkey. ► Our models can be easily used by province managers and decision makers for future planning.
Introduction
World energy demand has enormously increased since many countries required primary energy sources for sustainable development. For environment concern, natural gas is desired more among other primary energy sources. The primary use of natural gas is not only for heating but also generating electricity in thermal plants. This makes it crucial contribution to a country in term of economic, social and technological developments. This results in determining natural gas demand in order to achieve proper energy management policy for local decision makers in some countries including Turkey, Romania, Bulgaria, etc. It is fact that Turkey's fast growing economy is in process to integrate into the European Union's economy and to achieve its economic growth to compete with some of countries in the EU. This simply leads to an increase in Turkey's energy demand but Turkey's natural energy sources are unfortunately insufficient to meet this demand. Therefore, major part of primary energy sources including oil, coal, natural gas, etc., are inevitably imported from Russia, Iran, Algeria and Nigeria.
Turkey has begun to use natural gas for heating since 1987 and yearly natural gas consumption has reached 30.9 × 103 m3 since then. 18.6%, 21.4% and 59.8% of natural gas amount are consumed in industry, residential homes and power plants. It is highly interesting that 45% of Turkey's electricity generation is obtained from natural gas cycle plant and this amount is surprisingly greater than the average world natural gas consumption. Major part of natural gas is consumed by residential places and industrial applications in Turkey's heavily populated cities located in Marmara region. This region accommodates foremost industrial corporations hence 7.3% of Turkey's generated electricity is consumed in this region. Although natural gas consumption is higher in this region, current literature survey indicated that there is still a limited number of study on forecasting natural gas consumption in regional basis. Furthermore, Turkey must consume a certain amount of natural gas according to some agreements signed between Turkey and the related countries. Therefore, it is usually encouraged to create new investments on natural gas consumption. These investments should be carefully planned and managed to minimize economic losses by selecting the appropriate forecasting models.
It is also necessary to precisely predict amount of natural gas consumption to eliminate undesired circumstances including weather conditions, regional development standard, etc. Moreover, accurate natural gas forecasting helps local administration to obtain necessary information on building infrastructure, minimizing air pollution and maintaining clean environment.
The aim of this study is to forecast short-term natural gas consumption in Sakarya province situated in north east Marmara region using the multilayer perceptron ANNs with time series. Sakarya province having important industrial corporations has grown its population by taking many immigrants from every part of the country. Natural gas has been consuming in this province since 2000 hence with new investments on infrastructure the number of customer has increased more than before. It should be emphasized that the major part of natural gas is consumed in residential homes rather than industrial corporations and there is a natural gas plant generating 19 billion kWh of energy in this province. The amount of this energy corresponds to 10% of Turkey's total energy generation from all of the power plants. In order to determine natural gas consumption trend for this province the seasonal time series with additional variables (SARIMAX), multilayer perceptron ANNs and RBF models are employed. The data used for these models cover daily natural gas consumption and as some meteorological data such as humidity, atmospheric pressure, wind speed, ambient temperature and average cloud cover in 4 years (2007–2011). It is believed that this is very helpful for the province administration in short-term energy investment plans. The remainder of this paper is organized as fellows: Literature review about natural gas forecasting is presented in Section 2. Descriptions of the proposed approaches are given in Section 3. In Section 4, the proposed models results are given to forecast the natural gas consumption in Sakarya. The last section is the conclusion of the paper.
Section snippets
Literature review
There are a limited number of studies on short-term-regional natural gas demand forecasting models in regional basis using daily based data. Among these models, statistical and stochastic forecasting methods are employed for short-term energy estimation in few countries. Yumurtaci and Asmaz [1], Akay and Atak [2] proposed an approach based on the gray prediction with rolling mechanism to predict Turkey's total and industrial energy consumption. Ceylan et al. [3] developed an approach to predict
Methodology
In this study, the daily natural gas consumption was modeled by using SARIMAX model, ANN-MLP and ANN-RBF models utilizing some meteorological data. Incomplete meteorological data were completed by local regression techniques and then statistically analyzed to preprocess. Later, the most appropriate time series model was chosen by using statistical methods on stationary data and parameters of SARIMAX model were estimated. Similarly, structure of ANN-MLP and ANN-RBF models were designed according
Results and discussions
In this study, the daily natural gas consumption was estimated by various models based on seasonal multivariate time series, ANN with MLP, ANN with RBF and multivariate OLS. Time series modeling part of the study on stationary series is resulted in using the model SARIMAX (1,1,0)(0,1,1)7. Maximum log-likelihood estimation of non-seasonal (p,q) and seasonal (P,Q) AR and MA coefficients for SARIMAX model are tabulated in Table 3. The statistical results showed that AT and CC were only
Conclusions
The principle purpose of natural gas consumption for residences is heating; hence meteorological factors affect gas demand and also consumption rates basically. Residential natural gas consumption gradually increases as Turkey's economic growth does. Gas consumption rate exhibits a sharp increase due to a rise in the number of natural gas consumers with respect to those in previous years in Turkey. Four-year daily data set seasonal pattern shows that gas consumption increases in winter as it
References (32)
- et al.
Grey prediction with rolling mechanism for electricity demand forecasting of Turkey
Energy
(2007) - et al.
Transport energy modeling with meta-heuristic harmony search algorithm, an application to Turkey
Energy Policy
(2008) - et al.
Electricity estimation using genetic algorithm approach: a case study of Turkey
Energy
(2005) - et al.
Residential–commercial energy input estimation based on genetic algorithm approaches: an application of Turkey
Energy and Buildings
(2004) - et al.
Modeling and prediction of Turkey's electricity consumption using artificial neural networks
Energy Conversion and Management
(2009) Turkey's natural gas necessity, consumption and future perspectives
Energy Policy
(2006)- et al.
An application of the degree-hours method to estimate the residential heating energy requirement and fuel consumption in Istanbul
Energy
(2000) - et al.
The degree-day method to estimate the residential heating natural gas consumption in Turkey: a case study
Energy
(2003) Natural gas demand in Turkey
Applied Energy
(2010)- et al.
Forecasting the growth of China's natural gas consumption
Energy
(2011)
Stochastic analysis of natural gas consumption in residential and commercial buildings
Energy and Buildings
Modeling and forecasting natural gas demand in Bangladesh
Energy Policy
Use of artificial neural networks for transport energy demand modeling
Energy Policy
Electric energy demand of turkey for the year 2050
Energy Sources
Ant colony optimization approach to estimate energy demand of Turkey
Energy Policy
Modelling of gas demand using degree day concept: case study of Ankara
Energy Sources
Cited by (111)
The detection of unaccounted natural gas consumption: A neural networks and subscriber-based solution
2024, Engineering Science and Technology, an International JournalApplication of forecasting strategies and techniques to natural gas consumption: A comprehensive review and comparative study
2024, Engineering Applications of Artificial IntelligenceAn intelligent framework to assess and improve operating room performance considering ergonomics
2023, Expert Systems with Applications