Content uploaded by Saud M. Al-Fattah
Author content
All content in this area was uploaded by Saud M. Al-Fattah on Mar 08, 2014
Content may be subject to copyright.
A preview of the PDF is not available
Time Series Modeling for U.S. Natural Gas Forecasting
Abstract
Faced with diminishing supplies of domestic crude oil and increased demand for energy, the US has come to rely on imported crude and domestic supplies of natural gas. During the last decade US natural gas production has risen about 10%, but it still fails to meet demand. This failure has resulted in significant increases in natural gas prices. We believe that developing a reliable method to forecast US natural gas production rates and reserves will benefit gas producers, consumers and policy makers.
This paper presents one methodology for developing forecasting models for predicting U.S. natural gas production, proved reserves, and annual depletion to year 2025 using a stochastic (time series) modeling approach. The methodology is not mechanistic. A mechanistic model would examine individual geologic settings, exploration success, the physics of gas production and the rate of exploitation for provinces, basins, and reservoirs. However, to do so would result in an extraordinarily massive model that would be difficult, if not impossible, to develop and use. Instead we used a simpler approach which takes advantage of established trends in easily obtained published data.
Various time series models were tested and validated using data that are not used in the mathematical development of the models. Having adequately validated these time series models using historical data we believe that they can be used to make at least short time forecasts. Comparison of results of this study with other published forecast is also presented. Our forecasts show that U.S. gas production rate will maintain a production plateau of 18.7 Tcf/yr from 2005 to 2008 after which gas production will increase gradually from 19.0 Tcf/yr in 2010 to reach 22.5 Tcf/yr in 2025. We predict that U.S. gas production will have an average annual rate of increase of 0.5% from 2005 to 2015 after which the average annual rate of change of production will increase to 1.2% for the period of 2015 to 2025. Our forecasts also show that U.S. gas depletion rate will increase from 10.6%/yr in 2005 to 13.4%/yr in 2025. While the reserves discovery rate increases, U.S. gas proved reserves are predicted to increase from 197 Tcf in 2005 to 215 Tcf in 2010. Afterwards, the gas proved reserves will increase at an annual rate of 1.3% and are expected to reach 263 Tcf in 2025.
... linearidade) devem ser consideradas. Modelos populares para esse fim, especificamente na indústria de petróleo, são o Autoregressive Integrated Moving Average (ARIMA) [Box 1970;Fan et al. 2021;Al-Fattah et al. 2005] e redes recorrentes profundas, em particular o LSTM (Long-Short Term Memory) [Fan et al. 2021;Liu et al. 2020;Sagheer and Kotb 2019;Hochreiter and Schmidhuber 1997]. Esses modelos preveem várias etapas dentro de um horizonte de tempo. ...
Monitoring and forecasting oil and gas (O\&G) production is essential to extend the life of a well and increase reservoirs' productivity. Popular models for O\&G time series are ARIMA and LSTM recurrent networks, and tipically several lags are forecasted at once. LSTM models can deploy the recursive prediction strategy, which uses one prediction to make the next, or the multiple outputs (MO) strategy, which predicts a sequence of values in a single shot. This work assesses ARIMA and LSTM models for the forecasting of petroleum production time series. We use time series of pressure and gas/oil flow from actual wells with distinct properties, for which we developed predictive models considering different time horizons. For the LSTM models, we deploy both the recursive and MO strategies. Our comparison revealed the superiority of LSTM models in general, and MO-based models for longer time intervals.
... MSE can evaluate the differences between actual values (y true ) and predicted values (y pred ), represented as Equation (13), which can be used to measure the overall performance of the trained model and help reduce forecasting error. Compared with MSE, RMSE represents the average error of a single sample, which helps to understand the physical significance, as defined in Equation (14). MAE gives a direct measure of the difference between predicted outcomes and true values, as defined in Equation (15). ...
Production prediction plays an important role in decision making, development planning, and economic evaluation during the exploration and development period. However, applying traditional methods for production forecasting of newly developed wells in the conglomerate reservoir is restricted by limited historical data, complex fracture propagation, and frequent operational changes. This study proposed a Gated Recurrent Unit (GRU) neural network-based model to achieve batch production forecasting in M conglomerate reservoir of China, which tackles the limitations of traditional decline curve analysis and conventional time-series prediction methods. The model is trained by four features of production rate, tubing pressure (TP), choke size (CS), and shut-in period (SI) from 70 multistage hydraulic fractured horizontal wells. Firstly, a comprehensive data preprocessing is implemented, including excluding unfit wells, data screening, feature selection, partitioning data set, z-score normalization, and format conversion. Then, the four-feature model is compared with the model considering production only, and it is found that with frequent oilfield operations changes, the four-feature model could accurately capture the complex variance pattern of production rate. Further, Random Forest (RF) is employed to optimize the prediction results of GRU. For a fair evaluation, the performance of the proposed model is compared with that of simple Recurrent Neural Network (RNN) and Long Short-Term Memory (LSTM) neural network. The results show that the proposed approach outperforms the others in prediction accuracy and generalization ability. It is worth mentioning that under the guidance of continuous learning, the GRU model can be updated as soon as more wells become available.
... Using the data from 1970 to 2007, they made a prediction study using a regression model. [16] analyzed the results based on historical data using the time series model for the US natural gas production forecast. ...
Technological advancements coupled with growing world population require the increasing need of energy. Natural gas is one of the most important usable energy resources. Turkey is with high external dependency on energy as it has its own limited natural and underground energy resources. Thus, in order to effectively and productively use of natural gas purchased from foreign countries and to make reliable and robust energy policies for the years ahead, it is crucial to make a reasonable and plausible prediction for natural gas consumption of Turkey. In this paper, we estimate the natural gas consumption using machine learning techniques on the basis of real monthly data representing natural gas consumption of Turkey between the years 2010 and 2018. The performances of machine learning techniques involving Artificial Neural Networks, Random Forest Tree, Regression, Time Series and Multiple Seasonality Time Series are compared in predicting the natural gas consumption of Turkey. Experimental results show that among the five techniques, artificial neural networks produce the best estimation, having the lowest mean square errors, followed by regression method. Time series shows the worst performance among all the techniques.
... According to Yeboah et al. (2012), various works on forecasting are based on the use of ARIMA models since the models produce accurate forecast values. The findings are found in various empirical works (Wang and Meng, 2012; Ahmad and Latif, 2011; Albayrak, 2010; Mucuk and Uysal, 2009; Erdoğdu, 2007; Al-Fattah, 2006; Ediger et al., 2006). For comprehensive review of these works, refer to Yeboah et al. (2012). ...
For a sustainable economic development, premium fuel forecasting is becoming increasingly relevant to policy makers and consumers. The current paper develops a structural econometric model of premium fuel using the Autoregressive Integrated Moving Average (ARIMA) to analyse and forecast premium demand. The results show that the ARIMA models (1, 1, 0); (0, 1, 1) and (1, 1, 1) are the appropriate identified order. The estimated models included a constant term. All the coefficients of the variables in the model except the constant term were significant. The diagnostic checking of the estimated model shows ARIMA (1, 1, 1) as the best fitted model since all the series were randomly distributed. The data for the forecast covers the period 2000:01 to 2011:12. The results indicated that the forecasted values fitted the actual consumption of the energy variables since the forecasted values insignificantly underestimate the actual consumption and thus indicate consistency of the results. The evaluation statistics indicate that the estimated models are suitable for forecasting. The model developed in the work is helpful to the energy sector and policy makers in making energy related decisions and investigating the changes in premium demand.
Keywords: Premium fuel; ARIMA; Forecasting
JEL Classifications: C51; C52; C53; E17; Q47
... Some of these models are qualitative and quantitative (see Yeboah & Ohene-Manu, 2015). The autoregressive integrated moving average (ARIMA) is one of the popular forecasting models that have been used in producing accurate forecast (As'ad, 2012; Yeboah, Ohene-Manu, & Wereku, 2012; Wang & Meng, 2012; Ahmad & Latif, 2011; Albayrak, 2010; Kumar, Kumara, Mallik, & Shuklaa, 2009, Mucuk & Uysal, 2009 Erdoğdu, 2007; Al-Fattah, 2006; Lloret, Lleonart, & Sole, 2000 ). For example, Yeboah and Manu (2015) used the ARIMA model to produce accurate premium forecast for Ghana. ...
The aim of the paper is to contribute to the body of knowledge in the area of forecasting using Autoregressive Integrated Moving Average (ARIMA) modelling for liquefied petroleum gas (LPG) for Ghana using monthly data for the period 2000-2011. The ARIMA (1, 1, 1) model was identified as suitable model. The findings show that the forecasted values insignificantly underestimate the actual consumption and thus indicate consistency of the results. The values of the evaluation statistics such as the ME; MSE; RMSE; MAE, and Theil's statistic, on the accuracy of the model indicate that the estimated model is suitable for forecasting LPG. The findings support the continuous use of the ARIMA model in forecasting, in econometric time series forecast. Future study should consider modelling other energy sources that are used in Ghana and other developing economies such as kerosene.
... We must reduce gas consumption rate to delay gas peaking. American shale gas reduced natural gas demand in the world market (Al-Fattah, 2005). Global ultimate natural gas reserves estimate may vary from 9500 to 15,400 TCF. ...
Fossil fuels production peaks, declines and depletions depend on their proved reserves, exploration and consumption rates. Worldwide proven oil, gas and coal reserves are 1,688 Billion barrels (Bb), 6,558 Trillion Cubic Feet (TCF) and 891 Billion tons (Bt) being consumed at rates of 0.092 Bb, 0.329 TCF and 7.89 BT per day, respectively. The oil, gas and coal reserves are increasing at the rate of 600 Million barrels (Mb), 400 Billion Cubic Feet (BCF) and 19.2 Giga Tons of Oil Equivalents (GTOE) per year. While the rate of annual increase in consumption of oil, gas and coal is 1.4Mb, 4.5BCF and 3.1 Million tons (Mt). Global annual energy demand of over 12 Billion Tons of Oil Equivalent (BTOE) results in the emission of 39.5 Giga tons of carbon dioxide (Gt-CO2), and the annual CO2 emission would increase up to 75 Gt-CO2 when future energy demand will rise to 24 to 25 BTOE. Oil, gas and coal may continue to exist for next several decades, yet the energy transition to low carbon intensity fuels is necessary to cope with rampant climate changes. Renewable and alternative energy sources hold key to the solution of twin problems, energy and climate change, with a high initial investment. Transition from fossil fuels to sustainable and renewable energy resources of 150 Petawatt hours (PWh) requires major investment and innovatory technologies. Perhaps CO2 and H2O based fuel systems would facilitate climate change and grand energy transition. An energy mix consisting of fossil fuels, hydrogen, bio-fuels, and renewable energy sources seems to be a good initiative. This paper reviews evidence of hydrocarbons decline scenarios and timelines of future energy technologies
Due to economic uncertainty and the financial crisis of 2008, a desire for an unregulated currency arose, leading to the invention of Bitcoin. Using a pseudonym called Satoshi Nakamoto, Bitcoin was created in 2009, anonymously or by a group of unknown individuals. Since Bitcoin has been the most valuable cryptocurrency in recent years, its prices have fluctuated dramatically, making it difficult to predict their prices. Investors, businesses, risk managers, and market analysts can all benefit from being able to predict Bitcoin prices. By using the Bitcoin transaction data obtained from the Bitstamp website in this study, several different Machine Learning models are employed to determine the most accurate model for predicting Bitcoin prices. These models are based on 1-minute interval exchange rates in USD from January 1, 2012, to January 8, 2022. Analysis was performed primarily with Python, but it was also used and Hadoop, a distributed data storage and processing framework that uses the map-reduce programming model to allow efficient parallel processing of Big Data. Based on the results of our research, comprising three experiments, autoregres-sive-integrated moving average (ARIMA) makes the most accurate prediction of Bitcoin prices, with a 95.98% success rate.
Nowadays, huge quantities of coalbed methane (CMB) well production data have been saved in the fields’ database, which declares that the industry of CBM enters the age of Big Data. The traditional gas production data analysis methods, such as the decline curve and type curve, are not effective for complex production conditions. So, we proposed an automatic segmentation trend tracking Model for the CBM production forecast, which is based on the production data and is more robust for the complex production condition.
Natural gas stands out among fossil fuels because it is relatively cleaner. It is also an important energy source type for several fields such as electricity production, industry, and heating, etc. Due to the poor capacity of Turkey in terms of natural gas sources, the demand is supplied by producer countries. Hence, accurate forecasting for the demand is of critical importance for Turkey, which imports 99% of its natural gas consumption. In the current literature about demand forecasting, most studies were conducted on an annual basis. However, the seasonal effect on the demand for natural gas cannot be foreseen through annual studies. Besides, to deal with some situations such as seasonal balancing, peak shaving, and gas supply shortage in monthly demand, forecasting models that capture the seasonal trend are needed. Therefore, in this study, a new grey seasonal forecast model has been presented and Turkey’s monthly natural gas demand was predicted via the proposed model. Performance of that model was compared with SGM(1,1) and SARIMA (p,d,q) x (P,D,Q)s. The obtained results show the superiority of the proposed model. By using this model, Turkey’s monthly natural gas demand was forecasted up until the year 2025. The proposed model allows us to capture seasonal patterns more successfully. In case this seasonal behavior continues, Turkey’s natural gas demand is expected to increase by %20 until 2025. At this point, the outcomes of the study provide important information to decision-makers to be able to determine reliable and stable energy policies.
In this chapter, tools from univariate time series analysis and forecasting are presented and applied. Time series components, such as trend and seasonality are introduced and discussed, while time series methods are analyzed based on the type of the time series components. In the literature, linear methods are the most commonly used. However, real time series data often include nonlinear components, so linear time series forecasting may not be the optimal choice. Therefore, also a basic nonlinear forecasting method is presented. The necessity of these methods to logistics service providers and 3PL companies is presented by case studies that present how the operational and management costs can be cut down in order to ensure a service level. Short term forecasts are useful in all the units of activation of 3PL companies, i.e. supplies, production, distribution, storage, transportation, and service of customers.
World gas supply forecasting has proved difficult because its exploration, transportation, and customer bases depend so heavily on fluctuating economic factors. Our recent study showed that the conventional Hubbert model with one complete production cycle is not appropriate to use to forecast gas-production trends for most gas-producing countries.
This paper presents our forecast for the world’s supply of conventional natural gas to Year 2050. We developed a “multicyclic Hubbert” approach that accurately models the gas-production history of each gas-producing country. Models for all countries were then used to forecast future production of natural gas worldwide. We present the multicyclic modeling approach in a convenient form that makes production data that exhibit two or more cycles easier to model and aggregated these models to regional and world levels. We also developed and analyzed supply models for some organizations [e.g., the Organization of Petroleum Exporting Countries (OPEC), the Organization for Economic Cooperation and Development (OECD), the E u ropean Union (EU), and the Intl. Energy Agency (IEA)]. Our results indicate that the world supply of natural gas will peak with a plateau production of 99 Tcf/yr from 2014 to 2017, followed by an annual depletion rate of 1%/yr. Regional analyses indicate that gas production of some regions will peak soon and that North American gas production is now (1999) at its peak. West European gas production is predicted to peak in 2002. Former Soviet Union (FSU) and Middle East countries, which contain appro x imately
60% of the world’s ultimate recoverable natural gas, will be the main sources of supply in the future .
Annual Energy Outlook DOE/EIA-0383, Office of Integrated Analysis and Forecasting
- Energy Information Administration
Energy Information Administration: "Annual Energy Outlook
2005," DOE/EIA-0383, Office of Integrated Analysis and
Forecasting, U.S. Dept. of Energy, Washington, DC (February
2005).
International Energy Annual
- Energy Information Administration
Energy Information Administration: "International Energy Annual
2001," DOE/EIA-0219, Office of Energy Markets and End Use,
U.S. Dept. of Energy, Washington, DC (March 2003).