SCHEDULE DELAYS AND NEW FINANCING FOR THE ARGENTINE ELECTRICITY
SECTOR GROWTH ,
J. Rego
National Council of Scientific and Technological
Research of Argentina - CEMA
Virrey del Pino 3210
(1426) Capital Federal ~ ARGENTINA
ABSTRACT
Apparently the electricity sector of Argentina has suffered
in the past from excessive capacity planning, owing to overoptimistic
forecasts of the demand growth rate. But because of the long delays
involved and lack of financial backing this advantage has been
progresively lost. The present official planning only provides for
renovation and demographic growth, not allowing for economic growth.
Therefore, the actual supply-demand balance of electricity can easily
be worn away by technical obsolescence and aging process of the actual
installed capacity of electricity production.
The problem behaviour arises when the timing of new capacity
investment is delayed, falling behind the programmed schedule of new
plants, without being able to meet the electricity demand . This
could happen mainly due to political prices well below costs because
of the inflation and or social subsidization, which leads, in turn, to
the discapitalization of the sector, that still remains nationalized.
A system dynamics model is used to explore the trade-off between
construction delays (which entails costs of unsatisfied demand) and
construction speed up (which entails financial costs).
Introduction.
The prevalent opinion in official circles -shared by some
from the private sector - that the actual installed capacity of
electricity generation, about 12,000 Mw, was big enough to satisfy a
peak demand of 8,500/9,000 Mw, giving a sufficient safety marginto the
system» was corrected by the facts. The combination of unhappy
circumstances, like an extremely dry winter season during 1988 and the
reparation works on the Chocoy Dam, lowered the hydraulic capacity
generation, making clear that such hipothetical reserve did not
exist, as public opinion learnt during that winter (1). Furthermore,
the non reliable thermical conventional capacity, due to obsolescence
has been estimated in the range of 2,500/3,000 Mw (Pescarmona, 1988),
leaving the system well below the security limits. There are
aproximately 5,500 Mw in the construction pipeline, which will be
ready in 5 or 6 years time, taking just technical considerations into
account , without financial delays. Regrettably, the big projects
involved in this period, would not be ready before 1992. Which would
be the picture for the next 5-10 years? Assuming a 20 years life-time
435,
for the conventional thermical and nuclear plants, and 30 years for
the hydro-electrical utilities, it would be necessary to remove 500
Mu, and build up an equivalent quantity annually in order to maintain
the present capacity. There ought to add up others 360 Mw more, for
taking care of a 3% annual rate of demographic growth. This would
entail to build between 950 and 900 Mw annually, according to
Pescarmona's estimations (1988), which amount to 5,400 Mw for the next
6 years.
Definition of the system model and its formulation.
The capacity growth controlling loop, is represented in
Figure 1. The forecast demand is generated, based on current demand,
and compared with the expected capacity over planning horizon, which
would differ from the present capacity because additions and
withdrawals are expected to happen in- that period of time. The
difference between the forecast and the demanded capacity defines a
future capacity gap, which has to be filled with new capacity orders
(Rego, 1982).
Figure 1. Capacity Growth Controlling Loop and Pressure of the Current
Gap on the Construction Delays and Scrapping Rates.
et Set
Construckon Ra’
Starking # seat
S xp ant + Capacthy
Potore ne sit
~, <r ag Capac ty =
+
Trend Electricity a
Demand F
The accounting sector of the model, shown in Figure 2, is
extremely simple and keeps track of the costs incurred when generating
electricity. The operating cost are the area under the load duration
curve weighted at each time interval by the operating costs per unit
of energy output, and the output of each kind of power station
operating in such interval. The overall performance of the sector is
evaluated by its return on investment, comparing the assets value with
profit, for every interval simulation time.
436
Figure Accounting Sector.
Capital + Utilities 4
he NSS chy
Interest ook
\ Operati Install
Revenes Set et _ tinge i Copacity
+
Blectriciy + Voriable PN n
Pisa Operating peed peeling
f “ Onitcese
Ceneration SY
Re L-
Rate
gueat
The approach to the model formulation is one which, in spite
of knowing that the problem adressed in this paper is a short-term
one, on the 5-6 years horizon, the model should, however, be able to
handle short and long-term trade-offs. Also it was sought ‘simplicity
and economy in its formulation. To this purpose the programme was
written taking full advantage of the macro and arrays facilities
availables in DYNAMO Plus. It has the ability to handle repeated
structures of a system using array variables. Doing so, the modeller
has to define every variable of the repeated structure in a generic
way, using subscripts, as it has to be done in most computing
languages, when a variable, e.g.» is defined inside a repeating loop.
This facility is used for the formulation of the power generating
capacity, which varies according to the ageing stage and the kind of
utility that generates the electricity. The double subscripted
variable cap(stageskind) represents simultaneously the capacity of
every type - nuclear, hydro-electrical, oil and gas-fueled; and every
stage -capacity in construction pipeline, installed normal capacity,
and capacity operating beyond its economic life, that is a 3x5 array.
Ss Sry
eSiom ption
There are different modelling traditions for the planning of
the electricity generation, like dynamic programming, linear and non
linear programming or integration of the load duration curve. The
system dynamics model used and presented in this paper follows the
last alternative. The cheapest way to meet the demand at any point in
time is to run the stations with the lowest operating cost. The
system operator tabulates the power stations in ascending order of
marginal operating cost and loads and unloads the stations
sequentially as the demand rises and falls (merit-order operation).
For clarity the system is aggregated in five representative power
stations: they are nuclear, new fossil, hydro, old fossil and gas
turbines. The power demand varies throughout the day and throughout
437
the year. To simplify the calculation of operating costs the load
duration curve is constructed from the daily demand curve rearranging
ach load for each time interval to occur in descending order of
magnitude. Thus the load duration curve makes integration of casts
less difficult because it can be represented by simple functions.
This curve characterizes the fraction of time the electrical load
is equal to or greater than a given output level. For example, an x
percent on the horizontal axis indicates that the load was u mw or
higher for x% of the year. The load duration curve is analytically
aproximated by an exponential function, where « is the parameter that
represents the rate of decreasing of the exponential aproximation to
the load curve. p is the peak power demand, m means the minimum power
demand and x is the fraction of the year (Ford, 1982).
(L-x)
u=tp-men™ Ae mx Cee
uretu
In order to obtain the total annual demand, given by the
area under the load duration curve, it is convenient to express the
value of x duration as function of the load demanded, in Figure 3.
«
)
1
(1/0
xei-(uj/x)) )
ay yen
The integral of function x fu, is then
Integral Fea, rau, -1704, 27%) « 4 C140 +c
Figure 3. The Load Duration Curve and the Portion of The Electrical
Demand Evaluated by Dgen Macro (After Ford, 19682).
eres Load Duration
. wSwrve
ot
xa d-(hy
Hi
epost)
438
The Dgen macro gives the area under the load duration curve
between the two ordinates L and U and is a modified version of macro
Demgen presented by Ford (1982). The internal variable $b evaluates
the area over the minimum demand and bi variable calculates the base
load area, under the line of minimum demand. The present formulation
is based on the solution previously found to the integral of function
Fluid.
x
macro dgen(pym,l,usa)
a $11.k=max(max(1.k»m.k)-m.k,yle-37)
@ Sul. k=max(max(u.k»m.k)-m.ky1e-37)
a $x1.k=max(p.k-m.k,le-37)
@ $al.k=(1/a.k)
a $a2@.k=(1lta.k) /a.k
a $a3.k=(1+a.k)
a $b2.k=($u1.k—-(1/$x1.k¥*Sal 2k) ) H0.k* (exp ($a2.k*logn($ul.k))/$a3.k))
a $b3.k=($11.k-(1/$x1.k*#Gal.k) Hake (exp ($a2.k#logn($11.k))/$a3.k))
a $b.k=$b2.k-$b3.k
a SDL kemin(usk mek) -min( lk »mek)
a dgen.k=($b.k+$b1.k)#8.76
mend
Now it is possible to follow a merit order in the
electricity generation. The nuclear available capacity, which appears
at the head of the proposed rank, becomes one of the arguments of the
dgen macro, which evaluates the electricity generated (nuclear) at the
base of the load duration curve, that represents such nuclear
capacity. Once that capacity was full employed; a second band of
energy is generated by all the available new oil-fueled utilities and
so on.
The sort of questions adressed by this paper has normally
difficulties in formulating the initial conditions of the system, in
particular the amount of capacity still in construction pipeline. The
built-in macro delays availables in DYNAMO Plus have the convenient
ability to initialize all their internal levels by themselves, so that
the inflow rates and their delayed outflows are in equilibrium at the
beginning of the simulation (Richardson, Pugh; 1981). But this does
not happen in the Argentine case, so it is necessary to built a delyép
macro where it would be possible to assign the actual capacity in the
construction stage into the corresponding internal levels, according
to the construction stage of each kind of power generation. The
output Delayép is a sixth-order exponential material delay.
Policy Analysis.
To become a truly System Dynamics Model it should lose its
rigidity, giving the possibility to the system to improve its
self-control. Something can be said to this respect about the role of
the actual gap between actual capacity and demand capacity; this is
the one perceived by the clients of the system, and failure in closing
such gap is sanctioned by the public opinion, putting pressure on the
439
management. Construction delays would have to hurry up and = scrapping
would be delayed, as shows Figure 1.
There are now two gaps to monitor simultaneously: one
concerned with the long term investment policy, which is the future
gap» and another one related to the day to day operation of the
electric grid, this is the present capacity gap. From the modelling
point of view, it is worthwhile to. comment that the construction
time delay, usually a constant argument of the macro delay which was
called, becomes a variable itself in this model. Such construction
delay depends on the value which takes the present gap. In a similar
way the scrapping time is no longer a constant but a function of the
same gap. The current capacity gap relative to the power demand
ranges from -5% to 10% in the low demand scenario, and from 5% to 15%
in the normal scenario so that it is convenient to use -0.20 and 0.20
as the lower and the upper limits of the independent variable current
gap, used in the table functions representing the proposed policies
This relative gap becomes the argument of the corresponding table
functions, to pursue the policies above described. The shape of the
functions connecting both variables with the current gap appears in
figure 4. This figure presents two alternatives to the originals
S0-SO0, DO-DO for each function used in the base runs. Those
alternatives ranked by its agressiveness are S1-S1, S2-S2, DI-Di, and
De-be.
Figure 4. Scrapping Time and Construction Delay Table Functions.
TC004,088) SCHTHC, 6H)
io - ~2 ed y
ciuge tes to” tse tote ‘Change tablet fonews!” Ese to continie
As it can be seen in Figure 5, the acceleration of the
construction programme, which is an obvious solution to the deficit of
electricity supply, reduces such a shortage at the expense of the
financial position which becomes worse and worse as the construction
time is shortened. This, of course, compromises the viability of the
Proposed solution and this not mentioning technical dificulties
implied by the acceleration programme itself. One of the problems is
that the extra profit generated through the reduction of deficit
capacity is not big enough to cover the heavy capital expenses
incurred in the acceleration of the construction programme. Its
consequences are similar in both and high demand scenarios, with the
undesired effect in the low case that the system will have to cope
with an excessive capacity for a longer period of time.
The scrapping policy which consists in being accelerated
when there is capacity excess and being delayed when there is
shortage, has shown its lack of effectiveness, at least as it has been
modelled. Actually, the lengthening of the old capacity lifetime, only
reduces the investment in the new capacity in the high and low
scenario improving the financial performance in both cases. It must
be said that this happens because the capital cost of the ald capacity
is considerably lower than the new one. The results could change
dramatically when the costs of enlarging the lifetime are properly
taken into account which is something not done in this preliminary
version of the paper. Improvement of the model in such a direction is
under study. As it is this policy cannot reduce the deficit. Perhaps
what should be done is to introduce in the model a conversion rate
which translates old capacity into new capacity which is also
something being considered.
Figure 5. Supply-Demand Capacity Balance and Financial Performance
Speeding up the Construction in the High Demand Scenario.
——— PLI(6000, 18.63) ee
18.e3---—- a
en A
80,03 f
-TOP(B, 80,63)
18.03) = —-—-— ZEROL(D, , 88, e-3)
[DO RETURN
ON
15,e3| a ~N| Df iwvestMen
Ce «
12.09
2-3}
9008,
20.e-3 D2 ANAILABLE
<. A Di = cAPACITY
6008, ~~. DO
g
1988, 1993, 1998, 2003,
TIME
Conclusion.
The electricity demand in Argentina is expected to
dissatisfied for the next five years or so, if that grows at the f
per cent historical annual rate. Simulating with a prelimin:
version of a System Dynamics model of the electricity supply sec
show that if new financing is available, which is represented by
assumption that payments for capacity additions are done not dur
the construction process but simultaneously with the incorporation
the new utilities to the grid, there will be some relief to
financial position of the sector, shown by an improvement of
return on investment in the next five years. But such effect is
worn down by the payments. With regard to the power capacity defi
the model’s control policies achieve the capacity-supply balance af
five years in the base case run, Such time can be shortened if
capacity construction is speeded up, but at the expense of
financial position. The extra revenues coming from such advance
the schedule would not compensate the required capacity paymen
This is probably the reason why this would not happen, rather
contrary.
Notes.
1. "La Nacién" Journal, Buenos Aires, 15th August 1988, p. 1.
References.
Ford, A.; Youngblood, A. (1982): “Technical Documentation of the
Electric Utility Policy and Planning Analysis Model.
Version 4". Los Alamos National Laboratory. University
California. Paper LA-9347-M5.
Pescarmona; E. (1988): "El problema de la energia eléctrica en la
Argentina", "Ambito Financiero” Journal, Buenos Aires,
9th May 1988, p. 14.
Rego, J. (1982): “Long-Term Policies for Electrical Generating
Capacity in Argentina". Unpublished Ph.D. Thesis,
University of Bradford.
Richardson, G.3; Puch, J.(1981):"Introduction To System Dynamics
Modelling". Cambridge, Mass.: The MIT Press.
Secretaria de Energia (1986): "Plan Energético Nacional 1986-2000:
Resumen." Ministerio de Obras y Servicios Publicos de la
Republica Argentina.
Solanet, M. (1988): "Situacién y perspectivas del suministro
eléctrico", Energeia, Mensuario Energétice Argentino,
No. 65, August 1988, pp.24-26.
CC BY-NC-SA 4.0