Electricity demand projection
The foundation for the PDP is an official forecast of future electricity consumption. Because electricity cannot be cost-effectively stored at national-scales, supply must be balanced with demand at every moment. Thus, the forecasted peak demand figure is important because the peak demand (plus the agreed-upon reserve margin), determines the amount of power plants that are necessary to ensure adequate power supply in the country. Because power plants and other related investments have long lead-times (typical large thermal plant requires two to three years of construction time, a typical hydropower plant requires at least four, and nuclear power plants at least five not including licensing and approval[2]), planning ahead is necessary to avoid power shortages. However, inaccurate forecasts could also lead to either a shortage situation (too few power plants built) or surplus situation (too many power plants built). Each has significant economic ramifications. As described below, Thailand load forecasts have consistently led to expensive power plant surpluses.
The Thai Load Forecast Subcommittee[3], under the Ministry of Energy, makes projections of the country's future electricity demand and updates them approximately every two years, or when circumstances change.
The key features of the methodology used to forecast demand are as follows:
- demand projections are primarily based on medium and long-term GDP growth forecasts (Vernstrom 2005).
- a secondary source of information comprises end-use models for certain customer classes with sufficient available data (residential, and some commercial and industrial customer classes).
- the fundamental underlying concept is one of exponential growth (annual increases are higher as the base (total consumption) increases).
Figure 1 shows the actual peak demand in Thailand (solid red line) compared to every forecast used to develop government power development plans over the past 20 years. There is a clear systemic tendency to over-estimate actual demand for electricity.
Figure 1: Government's load forecasts made in previous years of peak electricity demand (in MW) are all considerably higher than actual demand (solid red line at bottom of graph).
A second feature worthy of note is the curve of the forecast. All official demand projections are based on an assumption of exponential growth, with an upward bending curve that gets steeper over time. This exponential shape arises because exponential GDP growth is the main underlying driver in the government's power demand forecast model (Vernstrom 2004, EPPO 2007).
GDP growth rates adopted by forecasters have proven to be overly optimistic. Whereas planners predicted a base-case of 5.0% annual GDP growth from 2007 to 2011, actual GDP growth has averaged only 2.8% (see Table 4) over this period. One senior Asian Development Bank (ADB) official noted in a conference on power sector planning in the Mekong region, “Thai GDP figures are a little bit political, and are more like a wish numbers.” Few politicians would be excited to announce GDP forecasts during their time in office that predict mediocre economic growth. These wish numbers unfortunately lead to expensive and impactful over-investments in power generation.
PDP 2007 forecast Case 2007 2008 2009 2010 2011 5-yr avg 2012 2013 2014 2015 2016 5-yr avg 4.0 4.5 4.7 4.5 4.5 4.4 4.8 5.0 5.0 5.3 5.3 5.1 4.8 5.0 5.2 5.0 5.0 5.0 5.3 5.5 5.5 5.8 5.8 5.6 5.0 5.5 5.7 5.5 5.5 5.4 5.8 6.0 6.0 6.3 6.3 6.1 5.0 2.5 -2.3 7.8 1.0* 2.8
- Bank of Thailand's estimate, as reported in Matichon newspaper on 4 February, 2012
Table 3: Projected and actual GDP growth in 2007 – 2011, as well as predicted growth in 2012-16. Data sources: (EPPO 2007; EPPO 2011;Yuvejwattana 2011).
Moreover, the problem with GDP forecasts is not all wishful thinking and politics. Part of the problem is that real growth of the Thai economy is affected by “Black Swan” events: unexpected occurrences such as the 1997 financial crisis, oil price spikes, violent political conflicts and a devastating “50-year” flood. These occurances were impossible to predict and were, of course, never taken into account in projecting future power demand. The fact is that while the disruptions have been different each time, significant disruptions have occurred time and again with significant impacts on the economy and electricity consumption (see Figure 2).
Figure 2: The growth of Thailand's gross domestic product (GDP) has not been without interruptions. The 1997 financial crisis, political instability in 2009 and "50-year" flood in 2011 each had significant negative impacts on the economy as well as power consumption.
Such unexpected events though hard to predict are part of the reality of the economy. Their effect has been to throw growth trajectory toward a path that has ended up being more linear or logistic shaped than exponentially growing. Given the on-going uncertainties of world economy, domestic political environment and extreme climate events, it is unrealistic to expect that forecast GDP and electricity demand will grow exponentially as predicted.
The extent of over-optimism in demand projection becomes more apparent when we compare the projected annual demand increase as assumed in the PDP2010 with the historical records, as shown in Figure 3.
Figure 3: Annual peak demand increase in MW: comparing the PDP2010 projections with the past actual records and historical averages. The difference between the PDP 2010 forecast growth and the highest actual historical average is about 660 MW in peak demand per year, equivalent to about one coal-fired power plant each year.
As shown above in Figure 3, when averaged over the past 25 years, demand for electricity in Thailand has grown about 830 MW per year. When averaged over 15 years, this shrinks to 813 MW per year. Over the past 10 years it demand has grown only 772 MW per year, and the past 5 years have seen an average increase of only 407 MW per year. In contrast, the PDP 2010 assumes average increase of 1491 MW per year. Considering the actual trend over the past 25 years (lower and lower increases on average every year), it is conservative[4] to assume that long-term future demand for electricity increase at the 25 year average rate.
In light of this track record, and the available data, the following revised assumptions are used in the demand forecast in the PDP 2012:
Assumption Rationale Actual 2011 peak used as base for projecting future demand The February 2010 PDP over-predicted 2011 peak demand 668 MW. Linear demand growth, based on historical 25-year average (830 MW/year) February 2010 forecast assumes uninterrupted exponential growth, whereas historical growth has been essentially linear (with declining averages in past years) and there are uncertainties, such as world economic condition and extreme climate events that affect the Thai economy. Table 4: Assumptions used in PDP 2012 demand forecast
Based on the above assumptions, the adjusted peak demand forecast, called “PDP2012 forecast” is as shown in Table 5. Compared to PDP2010 forecast, the PDP2012 peak demand for year 2030 is reduced by about 13,200 MW to 39,692 MW.
Table 5: Comparison of peak demand forecasts used in PDP2010 vs. PDP2012. Load Factor for the PDP 2010 and PDP 2012 are assumed to be identical.
Ultimately what is needed is for Thailand to move away from load forecasting based on econometric regression (top-down approach) and to invest instead in the capacity to undertake rigorous bottom-up forecasting that understands sector-by-sector, industry-by-industry, end use-by-end use what the actual growth in electricity consumption will be. This is data-intensive and requires much more detailed understanding of exactly how electricity is being used by all customer classes, and how these usage trends are affected by changing technology, appliance efficiency improvement rates, adoption rates, prices, domestic and international economic climate, and changing demographics. Though a formidable task, user surveys and data gathering and analysis are likely to be a much better investment than mistakenly building unneeded power plants.
[2] According to the French Nuclear Safety Authority (ASN), it takes at least five years to set up the legal and regulatory infrastructure for a nuclear power program, two to ten years to license a new plant, and about five years to build a power plant. That means a “minimum lead time of 15 years” before a new nuclear power plant can be started up in a country that does not already have the required infrastructure. Source: http://www.world-nuclear.org/info/inf102.html, accessed March 21, 2012.
[3] Chaired by Energy Permanent Secretary, the Load Forecast Subcommittee comprises mainly representatives from the three electric utilities, government agencies, large power users and a few academics.
[4] By “conservative” we mean assumptions that will lead to extremely low likelihood of insufficient power plants available to meet electricity load.
