Sunteți pe pagina 1din 8

204 IEEE Transactions on Power Systems, Vol. 5, No.

1, February 1990

A LARGE SCALE APPLICATION OF AN OPTIMAL DETERMINISTIC HYDROTHERMAL SCHEDULING ALGORITHM

A.A.F.M. Carneiro, Member IEEE and S. Soares P.S. Bond


Universidade Estadual de Campinas - UNICAMF' Universidade Federal do Paran; - UFPR
Faculdade de Engenharia Eldtrica - FEE Companhia Paranaense de Energia - COPEL
13081- -
Campinas SP 80000 -
Curitiba - PR
Brasil

ABSTRACT composite reservoirs in order to allow its


optimization by stochastic dynamic programming [1,2].
This paper presents an application of a This is the approach currently in use by Brazilian
deterministic optimization algorithm in the utilities in their operational planning methodology
hydrothermal scheduling of the large scale Brazilian [31*
south-southeast interconnected system, composed of 51 An alternative approach consists of the use of a
hydro and 12 thermal plants, corresponding to 45 GW of deterministic optimization tool where an optimal
installed capacity. The application considers the solution with individual hydro representation is
system operational conditions according to the 1986 obtained for forecasted inflows. This kind of approach
Operational Plan coordinated by ELETROBRAS the - must be implemented in an adaptive procedure [ 4 ] where
Brazilian electric holding company. The employed at each time period an updated solution is obtained for
algorithm is based on a network flow approach updated forecasted inflows.
especially developed for hydrothermal scheduling. For On the other hand, a deterministic optimization
the south-southeast interconnected system the problem tool can also be useful for operational planning
formulation suggests a primal decomposition studies when dealing with historical or synthetic
optimization approach. Two applications have been stream flows [5].
performed: (1) in a study context the system has been
evaluated for the most critical stream flow ever This paper is concerned with the use of a
recorded; (2) in an operational context the system has deterministic optimization algorithm [61 in the
been simulated through an adaptive operational planning Brazilian south-southeast interconnected hydrothermal
approach. The results show that the employed algorithm system (BSSIS), composed of 63 power plants
can be a useful tOQl for large scale hydrothermal corresponding to 45 ckl of installed capacity, and
systems, giving insight into operational responsible for more than 83 percent of the total
characteristics and/or adding to the guidance of real energy consumed in the country [3]. The 1986
system operation. Operational Plan [7] coordinated by ELETROBRAS - the
Brazilian electric holding company was used as a -
reference for system configuration and operational
INTRODUCTION conditions.
In an operational study context the (BSSIS) was
The objective of the energy operational planning evaluated under the most critical hydrological
of large scale hydrothermal systems is to obtain an conditions ever recorded. The results show the system
economic and reliable solution to guide real operation. performance and some interesting behavior
This is not an easy task due to several characteristics characteristics of the reservoirs.
of the problem: In a real operational context the (BSSIS)
- The necessity of a long horizon of analysis due to
the presence of large reservoirs of multi-year
operation was simulated by an adaptive planning
approach implemented in monthly steps, where the
regulation capacity. hydrothermal scheduling is optimized with inflows given
by a time series forecasting model [8]. The same year
* The stochasticity of water inflows. 1986 was used for this test and the results were
- The operative inter-dependence between hydroelectric
plants located in the same river basin.
compared with (1) the optimal solution obtained with
knowledge of the 1986 verified inflows, and (2) the
actual system operation in that year.
* The nonlinearity of thermal costs and hydro The paper is organized as follows: first, the
generation functions.
general deterministic formulation of the hydrothermal
These difficulties lead to the necessity of scheduling problem is presented; then, the (BSSIS) is
decomposing the operational planning into steps of described, according to the 1986 Operational Plan of
long, mid and short terms, constituting a planning ELETROBRAS, and its deterministic optimization is
chain where the particularities of each time horizon formulated and solved by a primal decomposition
result in the simplification of problem modeling. approach; then, the numerical results are shown for
The main issue in hydrothermal energy operational both the critical stream flow study and the adaptive
planning is the difficulty of adequately handling the planning simulation; finally, some points are commented
stochastic nature of inflows and the individualized and the final conclusions are stated.
representation of hydro power plants. A very common
approach has been to aggregate the hydro system into
THE HYDROTHERMAL. SCHEDULING DETERMINISTIC FORMULATION

The deterministic formulation of the hydrothermal


scheduling problem can be stated as the minimization of
the total operational cost given by
89 SM 673-5 PWRS A paper recommended and approved
by the IEEE Power System Engineering Committee of the m

IEEE Power Engineering Society for presentation at the


IEEE/PES 1989 Summer Meeting, Long Beach, California,
-
July 9 14, 1989. Manuscript submitted January 30, 1989;
made available for printing June 20, 1989. where
t time period (week or month)
0885-8950/90/0200-0204/$01.00 @ 1990 IEEE
205

T number of time periods THE BRAZILIAN SOUTH-SOUTHEAST INTERCONNECTED SYSTEM


ft(.) operational cost function
The Brazilian south-southeast interconnected sys-
C(.) non-hydro generation tem (BSSIS) is composed of 51 hydro plants with 31
reservoirs, as shown in Table 1, and 12 thermal units,
The non-hydro generation is composed of thermal corresponding to an installed hydro capacity of 41524
generation (coal, oil, gas, nuclear), non-contractual MW and thermal capacity of 3845 MW. Its largest river
imports and even load shortage, and is given by basin (the Para& river basin), located in the
southeast system, includes 32 hydro plants and 14
C(t) = D(t) - R(t) (2) reservoirs as shown in Figure 1.
where
O(.) load demand
total hydro generation
€I(.)

As the non-hydro sources must be employed in an TABLE 1


economical order according to increasing marginal
costs, the operational cost function ft(.) is a convex
increasing function of non-hydro generation [9], and Hydro plants in the (BSSIS)
depends on the time period t due to maintenance
schedules and non-hydro system expansions.
The total hydro generation is obtained from Number Name InstalledPower Storage Capacity
(Mw) (106~3)
(3)
1 EmborcacSo 1200 12521
where 2 Itumbiara 2080 12454
N
Oi( - )
number of hydro plants
hydro generation function 3 Cachoeira Dourada 439 -
4 sHo si&o 1680 5540
xi(.) water storage in reservoir i 5 Camargos 45 672
ui(.) water release from reservoir i 6 Itutinga 49 -
7 Furnas 1216 17217
The hydro generation is a function of water head 8 Peixoto 476 2500
andwaterrelease and isgiven, avoiding the indices, by 9 Estreito 1050 -
-
$(x.u) - K.[hl(x) - h2(u)l.min[u,qmax1 (4)
10
11
12
Jaguara
Volta Grande
Porto ColGmbia
426
380
320
-
-
where 13 Graminha 80 504
K constant taking into account gravity accel- 14 Euclides da Cunha 108 -
eration, water density, turbine-generator 15 A.S. Oliveira 33 -
efficiency and conversion factors. 16 Marimbondo 1488 5260
hl(.) up stream level as a function of storage. 17 Kgua Vermelha 1380 5169
18 Ilha Solteira 3240 12828
h2(.) down stream level as a function of release. Barra Bonita 2566
19 140
q- maximum water discharge. 20 A.S. Lima 143 -
21 Ibitinga 132 -
The dynamic water balance equations for each 22 Promissa0 264 2128
reservoir are given by 23 Nova Avanhandava 302 360
24 Trbs I d o s 810 3040
xi(t+1) = xi(t) + Yi(t) + 1 uk(t) - ui(t) (5) 25 Jupii 1414 -
kQi 26 Jurumirim 98 3165
27 Xavantes 414 3041
where
yi(t) incremental water inflow into reservoir i
28 L.N. Garcez 70 -
29 Capivara 640 5724
ni indexset of immediately upstream reservoirs 30 Taquarufu 500 230
31 Rosana 320 550
Equation (5) does not consider time lag between 32 Itaipu 12600 -
hydro plants since this is not necessary for the 33 Foz do Areia 1676 3805
operational planning of (BSSIS), in montly steps. 34 Santiago 1332 2670
However, if necessary, it can be considered without any 35 OzGrio 1050 425
difficulty . 36
37
PaCo Fundo
Paco Real
220
140
1387
3357
The storages and realeases are bounded due to 38 Jacui 180 -
operational limits, flood control, navigation, or other
purposes
39 Itauba 500 -
40 Parigot de Souza 252 156
41 Henry Borden 880 1311
42 Jaguari 28 783
43 Paraibuna 85 2636
44 Funil 216 606
45 Ilha dos Pombos 167 -
-
The solution of (1) (7) for a given initial 46 Fontes 158 450
-
condition xi(0), i=l,2,...N and given incremental water 47 Pereira Passos 100
inflows constitutes the deterministic individualized 48 Salto Grande 104 -
hydrothermal scheduling. Its solution can be carried 49 Mascarenhas 123 -
out by nonlinear programming techniques such as, for 50 Tr& Marias 396 152 78
example, the nonlinear network flow algorithm espe- 51 Nilo Pecanha 380 -
cially developed for hydrothermal scheduling [6].
206

2 - The hydrothermal maintenance scheduling over the


period.
3 - The interchange limits over the period.
4 - The storage constraints due to flood control.
/ .- 5 - The capacity scheduling of new hydro plants andthe
filling up of the minimal storage ofnewreservoirs.
6 - The initial and final reservoir storages.
The (BSSIS) deterministic formulation

The (BSSIS) deterministic optimization problem can


be formulated, according to (1)-(7), as the following P
problem
T
min J = 1 ~f:[c~(t)l + f:[c,(t)ll (8)
Fig. 1 - The Parani river basin t=l
s.t.
ci(t) = Dl(t) - Hl(t) - S(t) (9)
The inter-tie line between the south and southeast C2(t) = Dg(t) - H2(t) + s(t) (10)
subsystems has a tension level of 750/500 Kv. Its
transmission limits are different in each direction and -S(t) s S(t) s I(t) (11)
change Over the horizon due to system expansions and Hl(t) E Hi (12)
maintenance schedules, according to the Operational
Plan [7]. H2(t) E H, (13)
The operational cost functions for each subsystem where
have been calculated by optimizing the thermal
generation costs for each production level [9]. The S(t) energy interchange in the southeast-
corresponding marginal costs are shown in Figure 2. The south direction.
shortage cost, not represented in the figures, which I1,2) index set of southeast and south sub-
represents not a real cost but a penalty term, was systems.
considered to be a quadratic function with an average
cost of US$ 360 per MWh. the same adopted in [7]. -
S(t), S(t) energy interchange bounds over time.
Hi hydraulic constraints (3)-(7) for sub-
system i.

US $ /MW month US $/MW month The solution of the above problem was carried out
byanoptimal decomposition approach thatf-ixesat-each
iteration the feasible interchange S(t) 5 S(t) h S(t),

/
40 OOO solving the decomposed optimization subproblems
30-

s.t.
Ci(t) = Di(t) - Hi(t) + (-l)is(t) (15)
Hi(t) E Hi (16)

The optimal solution of (14)-(16) may present


different marginal costs between the subsystems. In
this case a new interchange level must be fixed by the
S WTHEAST MW month SOUTH MWmonth
coordinator in order to approximate the subsystem
marginal costs as near as possible to each other for
the current hydrothermal solution. Note that if the
Fig. 2 - Marginal costs in the (BSSIS) cost functions ft(.) are convex and differentiable the
marginal cost equalization will be assured for this new
interchange level. This is not the case for the (BSSIS)
system whose marginal cost functions shown in Figure 2
The numerical tests here presented were carried
presents many discontinuous points.
out using hydro generation functions given by (4) where
hi(.) and h2(.) are considered to bepolynomialfunc- The whole procedure is repeatedwhile it is possible
tions up to the fourth degree. todecreasethe marginal cost differences. This decompo-
sitionapproachcorrespondstoa projection of the origi-
The performed tests have been applied to the
(BSSIS) according to the 1986 Operational Plan nal problem P into the spaceoftheinterchangevariables.
developed by ELETROBRAS [7], which covers 5 years ahead
(1986-1990). Based on its statements the performed min {min P/s(t)l
deterministic optimization has considered: S(t) ECt)

1 - The global system configuration over the period NUMERICAL RESULTS


86-90, including the installed capacity expansion
and the load demand evolution. The main objective in the numerical experiments
201

here presented was to evaluate the (BSSIS) behavior The operational cost evolution of each subsystem
using a nonlinear network flow algorithm, previously for the first interchange level (S(t) = 0) is shown
developed [6]. In this way two applications have been in Figure 4.
carried out: (1) in a study context the system During all performed tests a run-of-river initial
behavior over 1986-1990 has been evaluated under the solution was adopted. This initial solution presents
most critical historical stream flow;(2) in an oper- many periodsof load shortagewith very high operational
ational context the system has been simulated during costs, whichjustify the unrealistic large initial costs
1986 ueing an adaptive planning approach. shown in Figures 3 and 4. ?he final solution gave
a total operational cost of 144 million dollars per
Critical stream flow study year, being 105 million dollars in the southeast sub-
system and 39 million dollars in the south subsystem.
The (BSSIS) optimization was performed for the The optimal interchange level obtained is shown in
1986-1990 configuration. considering the critical his- Figure 5. Note that the inter-tie transmission limit is
torical stream flow which occured in 1952-1956. The binding in 1988 and 1989 when the transmission capacity
results are presented below. in the direction south-southeast is almost zero. In the
Figure 3 shows the convergence of the interchange remaining period the interchangeoccurs in the direction
coordination which can be practically considered to southeast-south without achieving the limits.
have been achieved in two iterations. This behavior
occurred because the energy interchange represents a
small percentage of the total system load demand.

L, - OPTIMAL
I --- BOUNDS
I
aCf I

--- SOUTH
......... SOUfHE4ST
- GLOBAL

I y - 7_...................................................
-I 2

-20
Fig. 3 - Evolution of the operational costs in the
(BSSIS) Fig. 5 - Optimal interchange from south to southeast
subsystems

The south hydro generation is shown in Figure 6


and the southeast hydro generation i s shown in Figure
7. Comparing initial and final solutions it can be seen
that the optimization tries to equalize the hydro
- SOUTH
production along the 5 years constrained by
regulation capacity of the reservoirs. Comparing the
the

........ SOUTHEAST subsystems optimal solutions, note that they are com-
plementary for the critical stream flow.
The subsystems marginal operation costs are shown
in Figures 8 and 9. It can also be noted from other
point of view that the optimization tries to equalize
the marginal costs over time and between subsystems.
The optimal solutions with and without the
interconnection show that, for the critical stream flow
considered, only the south subsystem benefits with the
interconnection.
The storage trajectories of some important
................
-
I
, ................I ................. reservoirs are shown in Figures 10 and 11. The
Paranaiba river cascade composed of the reservoirs of
0 2 000 4 000 so00 the EmborcaCZo, Itumbiara and Si0 S i 6 0 hydro plants
IT€RATIONS
was selected in order to show the different behavior of
Fig. 4 - Evolution of the operational c o s t s in the each reservoir storage trajectory, according to its
isolated subsystems position in the cascade.
208

These characteristics can also be observed in Fig. 11


with the reservoir of the Furnas hydro plant in the
M W month x 10'
:: upper part of the cascade and the reservoir of Ilha
!! Solteira hydro plant in the lover part of the cascade.

GLOBAL OPTIMAL SOLUTION


RUN-,OF-RIVER INITIAL SOLUTION

'Ij ---- RUN- OF-RIVER SOLUTION i


I
!J - OPTIMAL SOLUTION I

I I I I
I
1986 I987 I988 I989 I990

Fig. 6 - Hydro generation in the south subsystem


l9g6 I 1987 I 1988 ' 1989 I 1990 I

Fig. 8 - Marginal operation cost in south subsystem


M Wmonth x IO3
---- RUN-OF-RIVER SOLUTION
- OPTIMAL SOWTION
n
US $ x I05/MWmonth

i i
I I

1986 ' 1987 I988 1989 l9SO


I986 1987 I988 1989 I990
Fig. 7 - Hydro generation in the southeast --- RUN-OF-RIVER INITIAL SOLUTION
subsystem ......... ISOLATED SUBSYSTEM SOLUTION
-GLOBAL OPTIMAL SOLUTION

The reservoir of the Ehborcafio hydro plant, which


is at the top of the cascade, decreases its storage Fig. 9 - Marginal operation cost in southeast subsystem
over the whole horizon, never reaching the top level
again. The reservoir of the Itumbjara hydro plant has
oscillating storage each year but reaches the top level This kind of behavior has been also observed for
in at least the three first years. The reservoir of the other stream flow sequences and indicates that each
SHo Si& hydro plant reaches the top every year and reservoir trajectory depends on its relative position
only oscillatesdue to floodcontrol storage constraints. in the cascade. For instance, the Ilha Solteira and S b
209

S G o hydro p l a n t s tend t o work as run-of-river p l a n t s , period. Once t h e v e r i f i e d inflows have occurred t h e


only decreasing t h e i r s t o r a g e l e v e l s due t o f l o o d procedure i s repeated t a k i n g i n t o account thenew system
c o n t r o l . In c a s e of t h e I l h a S o l t e i r a hydro p l a n t c o n d i t i o n s i n terms of c u r r e n t r e s e r v o i r s t o r a g e s and
t h e run-of-river c h a r a c t e r i s t i c i s due t o its high updated f o r e c a s t e d inflows. This kind of approach is
energy c o n t e n t s derived from i t s low p o s i t i o n i n t h e similar t o o t h e r suboptimal s o l u t i o n s f o r s t o c h a s t i c
cascade. In t h e case of t h e g o S G o hydro p l a n t t h e dynamic problems [4].
run-of-river c h a r a c t e r i s t i c i s a l s o due t o i t s low d i s - I n t h i s a p p l i c a t i o n t h e inflow f o r e c a s t s were
charge c a p a b i l i t y . On t h e o t h e r hand t h e r e s e r v o i r s of , obtained u s i n g a t i m e series model [ 8 ] a d j u s t e d over
t h e Fmborcwgo and F u r n a s h y d r o p l a n t s assume t h e r o l e o f f i f t y y e a r s of a v a i l a b l e h i s t o r i c a l data. The inflows
l a r g e multi-year energy r e g u l a t o r s i n t h e system, while were f o r e c a s t e d f o r t h e f i r s t 12 p e r i o d s only. For t h e
t h e i n t e r m e d i a t e hydro p l a n t s , such as Itumbiara, tend remainder of t h e horizon t h e long term average inflows
t o work as annual r e g u l a t o r s . were used s i n c e i n t h i s case f o r e c a s t i n g makes no
sense.
The (AP) s o l u t i o n i s compared with an optimal
s o l u t i o n t h a t considered t h e v e r i f i e d inflows of 1986
(OVI). These r e s u l t s a r e a l s o compared w i t h t h e a c t u a l
v e r i f i e d o p e r a t i o n of t h e BSSIS system during 1986
(AVO).
Figures 12, 13, 14 and 1 5 show t h e s t o r a g e trajec-
t o r i e s of some important r e s e r v o i r s of t h e cascade com-
posed by t h e Grande and P a r a n i r i v e r s , which are t h e
Furnas, hrimbondo, &ua Vermelha and I l h a S o l t e i r a
hydro p l a n t s . Note t h a t t h e (AP) t r a j e c t o r i e s t r y t o
f o l l o w t h e (OVI) t r a j e c t o r i e s , i n d i c a t i n g t h e a d a p t i v i t y
of t h e proposed approach. Eowever, as t h e adopted t i m e
p e r i o d w a s one month, t h e feedback is shown t o be rela-
ITUMBIARA ! .-, t i v e l y slow. A b e t t e r performance can be expected u s i n g
-S. S I M k '\J' a weekly d i s c r e t i z a t i o n .
0 I I '2/--q A -....-.y '
1386 1987 1988 ' 1989 ' I990

Fig. 10 - O p t i m a l storage t r a j e c t o r i e s 100L STORAGE (%I


-.
.-...... . -AVO

STORAGE (Oh)

20-
......: .
0-
J I F I M ' A ' M I J ' J ' A I S 1 0 ' N I D '

FURNAE Fig. 12 - Storage t r a j e c t o r i e s of Furnas power p l a n t

1986 1987 1988 1989 I990

Fig. 11 - Optimal storage t r a j e c t o r i e s

1 I :
These o p e r a t i o n a l c h a r a c t e r i s t i c s have n o t been
adequately considered by t h e methodologies i n use by
B r a z i l i a n u t i l i t i e s . These methodologies are based on - AVO \ ....
\'. 0 . -

composite r e s e r v o i r modeling and simulation t o o l s that .


.... .. AP
'-'./'..
u s u a l l y assume p a r a l l e l s t o r a g e f o r t h e r e s e r v o i r s . --- O V ' :\

The adaptive planning approach simulation

I n t h i s s e c t i o n t h e (BSSIS) has been simulated Fig. 1 3 - S t o r a g e t r a j e c t o r i e s o f Marimbondo power p l a n t


over 1986 i n a n o p e r a t i o n a l context using a n Adaptive
Planning (AP) approach.
The l a r g e d i f f e r e n c e presented by t h e (AVO)
The (AP) approach i s a sub-optimalsolution t o t h e t r a j e c t o r i e s with r e s p e c t ' t o t h e o t h e r two ones can be
s t o c h a s t i c v e r s i o n of problem (1) - (7) whose optimal p a r t i a l l y explained by s e v e r a l modification i n t h e
s o l u t i o n cannot be achieved by s t o c h a s t i c dynamic (SSSIS) c o n f i g u r a t i o n during 1986, of which t h e most
p r o g r a m i n g due t o problem dimensionality. important are:
The (AP) approach c o n s i s t s of optimizing problem
(1)-(7), a t e a c h t i m e p e r i o d , f o r f o r e c a s t e d inflows i n
o r d e r t o o b t a i n t h e r e l e a s e d e c i s i o n s f o r t h e c u r r e n t time
. a 10%l o a d shortage i n t h e south subsystem during t h e
f i r s t two months.
210

. t h e u n a v a i l a b i l i t y o f t h e Angra I nuclear p l a n t during


t h e e n t i r e year.
generation. This occurs because t h e mst economic
thermal u n i t i n t h e southeast subsystem during t h e
. at hte theunexpected load behavior during t h e year (lower
begining and higher a t the end of t h e year)
period, according t o t h e maintenance schedule, has a
marginal c o s t higher than t h a t of t h e south subsystem.
Note y e t t h a t t h e (Om) t r a j e c t o r y f o r t h e southeast
due t o t h e B r a z i l i a n government's economic plan
subsystem presents an i r r e g u l a r shape from May u n t i l
against i n f l a t i o n .
December, suggesting a non-optimal solution. However,
t h i s i s not t h e case because t h e southeast marginal
c o s t during a l l t h i s time i s constant, since only t h e
nuclear thermal p l a n t is being used (see Figure 2 ) .

YW month x I
O
'
- AVO
....... A P
6 --- OVL

..k. \
\

*r\.
I '. I
I
\
\

J ' F ' U ' A ' Y ' J I J ' A ' S ' O ' N ' D '

Fig. 14 - Storage t r a j e c t o r i e s of hgua Vermelha


power p l a n t

Fig. 16 - Interchangefromsouth t o southeast subsystems

01 L....
J."
' '
r -
J ' F ' M ' A ' Y J ' J A ' S ' O ' N ' D '
8 MW month a I
O'

:r--
Fig. 15 - Storage t r a j e c t o r i e s of I l h a S o l t e i r a
power p l a n t

However, i t must be observed t h a t t h e (AVO)


t r a j e c t o r i e s show c e r t a i n c h a r a c t e r i s t i c s which appear
...............
t o i n d i c a t e t h a t t h e whole system i s not working i n an
optimal manner. This can be seen, f o r example, i n
Figure 15. As has already been commented in the
operation planning s t u d i e s (see Figure l l ) , t h i s hydro
plant must operate as a run-of-river plant, i.e. it
- AVO ........
must decrease i t s storage l e v e l only f o r reasons of .......
flood control. Nevertheless, t h i s i s n o t its observed
(AVO) t r a j e c t o r y over the year. On t h e o t h e r hand, t h e
(AP) and (OVI) t r a j e c t o r i e s go together seeking to
maintain t h e maximum s t o r a g e l e v e l . The unexpected . .
0
behavior observed i n these t r a j e c t o r i e s f o r t h e f i r s t
two months is a consequence of t h e very high inflows
v e r i f i e d i n t h i s period ( t h e f i r s t t h r e e months a r e t h e
most wet ones) and inadequate i n i t i a l storage l e v e l s Fig. 17 - Thermal generation i n t h e south subsystems
f o r the reservoirs.
The optimal interchange l e v e l i s shared i n Figure
16. I n t h i s graph is very evident the d i f f e r e n c e General comments
between t h e (AVO) t r a j e c t o r y and t h e two o t h e r s , t h e
(AP) and (OVI) t r a j e c t o r i e s , i n d i c a t i n g t h a t the The t e s t s here reported have presented a CPU time
v e r i f i e d system operation was not optimized. which v a r i e s around 3 minutes i n an IBM 3090 computer.
Figures 1 7 and 1 8 present the thermal generations As t h e nonlinear optimization problem f o r t h e (BSSIS)
of south and southeast subsystems. Note t h a t , during system has 3840 variables, 1920 e q u a l i t y and 3960
t h e f i r s t four months, although t h e interchange limits i n e q u a l i t y c o n s t r a i n t s (the run-of-river hydro p l a n t s
a r e not a t t a i n e d , t h e southeast subsystem with no have been aggregated t o t h e upstream r e s e r v o i r s ) . this
thermal generation, does not i n c r e a s e t h e interchange CPU time has been considered resonable, and p e r f e c t l y
toward t h e south subsystem, which p r e s e n t s thermal s u i t a b l e f o r planning purposes.
21 1

REFERENCES

N.V. Arvanitidis and J. Rosing, "Composite repre-


- AVO sentation of a multireservoir hydroelectric power
.... ... AP system",'Transactions on Parer Apparatus and-Sys-
--- OVI -
tems, vol. PAS-89, nO 2 , pp. 319-326, February
1970.

N.V. Arvanitidis and J. Rosing, "Optimal operation


6 of multireservoir systems using- a composite repre-
,...... sentation", IEEE Gansact
_..____ions on Power Apparatus
.-- . ~ vol.
and -Svntems.~
~ . PAS-89, n9 2, pp. 327-335,
February 1970.

C. Ferreira, L.A.L. Barreto, T.A. Araripe Neto and


L.A.M. Fortunato, "Energy operation planning of
the Brazilian interconnected system", Study
Committee 39, CIGRE Meeting, Rio de Janeiro, 1986.
D.P. Bertsekas, "DynamicProgrammingand Stochastic
Control", Academic Press, New York, 1976.

R.E. Rosenthal, "A non-linear network flow


algorithm for maximization of benefits in a hydro-
Fig. 18 - Thermal generation in the southeast electric power systems", Operations Research, vol.
29, nO 4 , July-August 1981.
subsystem
M.F. Carvalho and S . Soares, "An efficient hydro-
thermal scheduling algorithm", IEEE Transactions
The period from January/l952 until December/l956 on Power Systems, vol. PWRS-2, n9 3, August 1987.
has been considered in the critical stream flow study,
according to the Operation Plan [7] which covers the ELETROBRAS GCOI, "Operation Plan 1986", Report
civil year. However, the first 4 months of this period SCEN-GTEN-SE-S-17/85 (in Portuguese).
corresponds to very high inflows since the critical
stream flow truly begins in, May/1952 (the hydraulic G.E.P. Box and G.M. Jenkins, "Time Series Analysis,
year). That is the reason for the singular behavior of Forecasting and Control", Holden-Day, Szo Francisco,
the reservoirs during these first months. 1976.
As the final reservoirs storages have been S . Soares, C. Lyra and H. Tavares, "Optimal gen-
considered free the reservoirs became empty in all the eration scheduling of hydrothermal power systems",
performed tests. Therefore, its behavior during the IEEE Transactions on Power Apparatus and Systems,
final periodofthe planninghorizoncannot be considered vol. PAS-99, pp. 1107-1115, 1980.
as a typical one.
The adaptiveplanningapproachhas shown a promising [lo] C. Lyra, 8. Tavares and S . Soares, "Modelling and
performance, especially if implemented in shorter optimizationofhydrothermalgenerationscheduling",
periods as, forexample, inweekly stages. Nevertheless, IEEE Transactions on Power Apparatus and Systems,
an exhaustive experiment with many different stream vol. PAS-103, pp. 2126-2133, 1984.
flowsmustbe carried out in order to precisely evaluate
its performance.

CONCLUSIONS Adriano A.F.M. Carneiro (M) received the B . S .


degree in electrical engineering from the Catholic
University of SHo Paulo and the M.S. in power systems
This paper has presented an application of a from the Federal School of Engineering of Itajubi. Mr.
deterministic algorithm with individualized repre- Carneiro is presently working at the State University
sentation of the hydro plants in the large scale of Campinas towards a Ph.D. degree in the area of
Brazilian south-southeast interconnected hydrothermal operation planning of power systems.
system.
The application has taken into account the
nonlinearity of thermal cost and hydro generation func- Secundino Soares received the B . S . degree in
tions, maintenance scheduling, flood control con- mechanical engineering from the Technical Institute of
straints,interchange limits and other hydro operational Aeronautic, the M.S. and Ph.D. degrees in electrical
constraints,accordingtotheOperation Plan coordinated engineering from the State University of Campinas. Mr.
by ELETROBRAS -
the Brazilian electric holding Soares is a professor at the System Engineering Depart-
company. ment of the Electrical Engineering Faculty at UNICAMP.
The application was performed in two different
context: (1) in a study context, the system has been Pericles S . Bond received his B . S . degree in civil
optimized for the most critical stream flow ever engineering from the Federal University of Parani and
recorded; (2) in an operational context, the system the M.S. degree in operation planning of power systems
has been simulated using an adaptive planning ap- from the State University of Campinas. Mr. Bond works
proach. at the Companhia Paranaense de Energia (COPEL), the
The results have shown that the algorithm can be electric energy utility of the state of Parani. and is
an useful tool for planning purposes in large scale an assistant professor at the Federal University of
hydrothermal systems. Paranii.

S-ar putea să vă placă și