Unformatted text preview: he key events in the area (rules of the game), the uncertainties and
what decisions can be taken, including the values. This is also based on welldefined scenarios leading to options and decisions. The overall strategy is to decompose a complicated problem into smaller
chunks that can be more-readily analysed and understood. These small pieces can be brought together to create an overall representation of the
decision situation. From this framework, the forecaster can take action and
maybe take a quality decision on the expected future sector area demands.
Decision analysis provides the options to model uncertainty through the
appropriate use of probability. Risk attitudes can also be incorporated into the
Scenario writing takes a well-defined set of assumptions and then develops
an imaginative conception of what the future would be like if these
assumptions were true. In this sense, scenarios are not future predictions in
themselves. Rather, they represent a number of possible alternatives, each
one based on certain assumptions and conditions. It is then up to the 45 decision-maker to assess the validity of the assumptions in deciding which
scenario is most likely to become a reality.
Kahn of the Hudson Institute did some work on scenario writing. Kahn (1964)
and Kahn et al. (1976) developed a number of alternative scenarios for the
world. In one scenario he assumed that an arms control agreement between
the United States and Russia would be reached and that China would follow
only a defensive policy, rather than an offensive policy. Based on these and
other assumptions, he developed a scenario describing a future socio-political
environment, following a predictable sequence of developments, constraints
and ideologies. From this set of assumptions, Kahn predicted a number of
scenarios. One was that Russia would lose control over the world communist
A book with the title “The Mind of the FOX” gives a very good description of
scenario planning in action. The basic idea can be described by the X and Y
Control 3 4 Options Uncertaintity Decisions 2 Certainty 1 Uncertainties Rules of the game Scenarios
Absence of Control Figure 3.2.1 - Graphical display scenarios in action 46 There is, for example, the possibility of a mine in an area A, and a smelter to
follow either in area A or in area B. To relate Figure 3.2.1 to the mine and
smelter example, Figure 3.2.1, is explained as:
1) Rules of the game: there are zinc deposits in area A that may lead
to mining activities in the area and a smelter in either Area A or B.
2) Uncertainty: When the mine will open and where the smelter will be
3) Options: There are mainly three options i) No mine, 0 MW increase
ii) Mine and smelter in the area, 140 MW increase iii) Mine, but
smelter in another area, 40 MW increase.
4) Decisions: Basically two decisions: when the mining activities will
start and where the smelter will be.
Decision trees and scenarios are useful techniques to structure the problem
and to reason through some key issues. The question of how much for each
year still remains however. The other difficulty is that annual load data only
exists for the past ten years and a forecast of twenty years is required. To overcome the problem, a non-linear program has been defined. A simple
S-curve is fitted on the ten actual loads. The objective function is to minimise
the sum of squares between the errors (actual minus forecasts) and the
predicted load for year twenty is set to the expected load for that year
(constraint). Large step load increases are added separately to the expected
The results of this method and the balancing algorithm are compared. The
differences are small. See Chapter 5 for the results. Willis uses a similar
method to trend the expected loads in an area. 
3.2.7 Transmission Substations
The transmission substation forecasts focus more on the networks, the
generation pattern and the end-use profiles. 47 A statistical software package, developed by EPRI (Load Dynamics), is used
to forecast the transmission substations’ future expected loads.
Two types of load forecasts are allowed. The user specifies the number of
years for the forecast. The maximum number of years is twenty years. In this
case the model generates a probability distribution on possible load levels for
each year, p(load|year). The model starts with the current load and growth
rate in the area. Alternatively the user specifies the largest load level that is to
be considered. This option is not used.
Next, the user is requested to specify the number of scenarios to be modelled
(2 – 5 scenarios). For each scenario the next inputs are required.
i) Label – to identify the scenario ii) Description – serves as a “comment field” to define the scenario iii) Scenario probability – the probability for the scenario to happen,
the sum of the probabilities for the different scenarios should be
one iv) Average Growth Rate over the Period (%) – self explanatory v) Lowest Growth Rate in any Year - self explanatory vi) Highest Growth Rate in any Year - self explanatory The next screen requests the estimated average number of years the next
three growth rates will persist. The growth rates are:
i) The lowest growth rate for the lowest scenario specifi...
View Full Document
- Spring '14
- Forecasting, Electricity generation, Electric power transmission, Power station, loads