The mean-squ are errors of each model with differen t
seasonal compon ents are tabula ted in Table 2. The MS
values are meaningful within the same item. Since these
numbers represent the goodness of each model within each
item, smaller the number better the model will be. The
results showed that seasonal order of 1 is suitable for each
item except hard coal, natural gas; solar, and total,
indicating the ARIMA model is suitable for them. On
the other hand, hard coal gives better model fit in seasonal
order of 3, natural gas in seasonal order of 2, solar energy
in seasonal order of 4, and total in seasonal order of 5,
indicating that the SARIMA model is suitable for them.
The ARIMA and SARIMA models constructed for each
item will not be shown here. Instead, as an example, the
result of forecasting for total primary energy is presented in
Fig. 1. The graph includes the realizations from 1950 to
2004 and the forecast and upper and lower limits of the
fifth-order SARIMA forecasting from 2005 to 2020.
The forecasted values for each item and for total from
2005 to 2020 is given in Table 3. As seen from the table, all
but wood and animal–plant remains will still be increasing
in the period from 2005 to 2020. However, the average
annual rates will decrease in all cases. For the period
between 1950 and 2004 the rate of increases are 8.3% for
oil, 7.3% for lignite, 4.3% for hard coal, 59.8% for natural
gas, 0.7% for wood, 0.2% for plant and animal remains,
18.3% for hydropower, 17.2% for petrocoke, 12.2% for
geothermal heat, 32.6% for geothermal electricity, 29.0%
for solar, and 259% for asphaltite and for the forecasted
period between 2005 and 2020 the rates are expected to be
1.6%, 1.3%, 5.1%, 6.8%, 1.2%, 1.8%, 1.7%, 4.2%,
2.2%, 3.1%, 6.6%, 1.4%, and 3.3%, respectively. On the
other hand, the average annual rate of total primary energy
demand decreases from 4.9% between 1950 and 2005 to
3.3% between 2005 and 2020.
Studying the relationship between energy and economy,
Ediger (2004) have shown that the industrialization in
Turkey has not been completed yet and energy demand
should be increasing faster than national income until the
energy intensity of the country reaches to a peak. Therefore, the decrease in the rate of energy demand may be
interpreted to indicate that the energy intensity peak will be
achieved in the coming decades. However, a close relationship exists between energy and economy of Turkey and the
average rate of change in GDP and primary energy
consumption are 4.5 and 4.9, respectively (Ediger and
Huvaz, 2006). Also, Soytas- and Sarı (2003) discovered
causality from energy consumption to GDP in Turkey.
Therefore, whether or not the decrease in energy consumption rate is related to energy intensity peak will
depend on the future rates of GDP. If causality runs from
energy consumption to GDP in the future and if the rates
of energy consumption and GDP persist their past trends,
any decrease in energy consumption is expected to slow
down the economic growth during the foreacsted period.
The numbers in Table 3 demonstrates that the share of
fossil fuels in sum of individual forecasts will increase from
87.6% in 2005 to 91.6% in 2020. The 4% increase in the
share of fossil fuels in total shows that the fossil fuels will
continue to be important in the future in Turkey similar to
most of the other countries in the world. However, the gradual increa se in the share of fossil fuels in total will
come from most ly the interacti on be tween oil and natural
gas in Tur key since the coal’s sha re in total fluc tuate s
betwe en 2 1.2% and 24.7% (Fig. 2). From 2005 to 2030, the
share of oil will decrease from 38.7% to 28.9% while that
of natural gas will increase from 25.7% to 41.1%,
indicating that 9.8% decrease in oil’s share will be
compensated by the 15.4% increase in natural Gas’s share.
The natural gas curve which, has already cut the coal curve
around 2000, is expected to cut the oil curve in 2012.
Similar results have previously been demonstrated in
electricity generation of Turkey by Ediger (2003b).
Additionally, the highest rate of increase will be recorded
in the future in natural gas. This indicates, although rate of
increase will decrease from 59.8% to 6.8%, natural gas will
continue to be a key element of the Turkish energy system
in the future. The cumulative graph of realized and
forecasted primary energy demand from 1950 to 2020
given in Fig. 3 demonstrates how important natural gas
demand will be in the future keeping in mind that it is met
almost completely by imports. In this context, the decrease
in both average rates and share in total in oil can be
considered reasonable since its domestic production meets
only about 10% of the demand. On the other hand, lignite,which is the richest energy source of the country, will
decreas e to only 1.3%. Therefor e, an interfuel substi tution
within fossi l fuels shou ld be mad e leadin g to a be st mix of
the cou ntry’s en ergy system. For inst ance, C- amdalı and
Ediger (in press) showed that a reduction of 1.663 billion
US $ in fossil fuel cost can be made possible by giving more
emphasis on domestic production particularly of oil,
lignite, and hard coal. Additionally, modern renewable
energy sources should be emphasized.
The ARIMA forecasting of total primary energy
consumption values and the summation of the ARIMA
and SARIMA forecasting of each item are also compared
(Fig. 4). As it is clearly seen in the figure, the ARIMA
forecasting of total primary energy consumption values are
higher than the summation of the ARIMA and SARIMA
forecasting of each item for the period between 2005 and
2010. It is just the reverse for the period between 2011 and
2020. The difference between the two is minimum in 2009
with a value of 83,000 toe and maximum in 2007 with a
value of 8.830 million toe. Considering that the ARIMA
forecasting is expected to give better results for short-terms
than long-terms and that during summmation the errors
involved in each forecasting are also summed up, the
ARIMA forecasting of the total primary energy demand
appears to be more reliable than the summation of the
individual forecasts.
Finally, the results of the ARIMA forecasting of the
total primary energy demand are compared with the most
recent MAED application (Table 4). It is obvious that in
the years 2006 and 2007 the low and high limits of this study almos t coinci des with the MAED resul ts. How ever,
for the long-te rm ARIMA foreca st gives an unde restima tion when co mpared wi th the MAED. The difference s
betwe en low er an d uppe r limits starts wi th 6.337 and 3.029
milli on toe in 20 08 and reach es to 58.324 and 6 9.736
milli on toe in 2020, respect ively.