To Main Proceedings Document |
OPERATIONS STRATEGY AND ENVIRONMENTAL MANAGEMENT IN COSTA
RICAN ELECTRICITY POWER SECTOR: A SYSTEM DYNAMICS APPROACH
ROY ZUNIGA
Permanent address: Barrio Cordoba, Urb. El Trebol, casa 11-C. Y Griega.
San J ose, Costa Rica.
e-mail: roy_zuniga@ hotmail
INTRODUCTION
By 1999, an almost totally integrated electricity power monopoly shows the Costa
Rican government responsible for 92% of total generation, 100% of transmission
and 82% of commercialisation. The performance level is, in general terms, good.
The electrification level is 93.25%, one of the highest in Latin America; electricity
losses are around 11%, one of the lowest in that region of the world; and the tariffs
are reasonable. However, future electricity generation and demand are important
concerns of the government due to: (a) the fact that between 35% and 47% of the
total government investment during last 10 years has been directed to this national
sector; (b) the increasing investment demands of other Costa Rican sectors; and (c)
the accumulated debt of the electricity sector is so big that during some years the
money required to pay the interests was bigger than the total income of the sector.
As a result, new legislation pursuing private investment at the generation level,
within certain limits, has been passed. The purpose of this paper is to show how not
only private investment will help Costa Rica to decrease its electricity debt, while
coping with the increasing requirements of the customers, but also how the
interaction of this action with the correct management of operational as well as
environmental issues will benefit the country.
There is a long tradition of System Dynamics models applied to the electricity power
industry (Aslam & Saeed, 1995; Barton & Bull, 1986; Bunn, Larsen & Vlahos, 1997;
Coyle, 1996; Dyner & Bunn, 1997; Ford, 1990a; Ford, 1990b; Ford, 1997a; Ford,
1997b; Ford & Bull, 1989; Ford, Bull & Naill, 1987; Grupo de Dinamica de Sistemas
de la Universidad de Sevilla, 1993; Larsen & Bunn, 1999; Lyneis, 1997; Naill, 1992),
1
which are very helpful in recognising several feedbacks present in this sector.
Unfortunately, this is insufficient to solve the specific case of Costa Rica. As Larsen
& Bunn (1999) have pointed out, each country faces different market and industry
structures, each holding different amounts of natural resources and generation
technologies. This situation leads to the combination and/or invention of models
suited to the needs of each country. The case of Costa Rica is no exception.
THE PURPOSE OF THE MODEL
The aim of the proposed model is to show, in an aggregated and strategic level, the
interrelationships among operations strategy issues (i.e., capacity, technology,
vertical integration, quality and production planning), environmental management
issues (i.e., conservation and efficiency programs, and losses management),
investment, debt, prices, cost, demand and forecasts; and how these variables
might interact with new regulation encouraging private power generation. This
paper extends the idea proposed by Pérez Rios (1999), who pointed out that the
turbulence of the environment stimulates the combined use of different
methodologies to tackle complex issues.
THE MODEL
Figure 1 shows the interrelationships among several strategic issues considered by
the model, from a highly aggregated perspective. For example, legislation and
customers influence the strategy of the electricity power sector, which influences its
operations strategy and its environmental management. Operations strategy is
practised through a number of decision categories including capacity, facilities,
production control and planning, technology, vertical integration and quality
(Wheelwright & Hayes, 1985). These decision categories are also influenced by
decisions related to environmental management issues. The interrelationships
among these decision categories will determine the level of performance and
competitiveness of the electricity power sector in terms of the well known distinctive
competences of price, quality, reliability and flexibility (Wheelwright, 1984).
Legislation Economic conditions Social considerations
AF GD
or: oN
sector re
Distinctive ae >.
“AN management
St)
co
Sie
a
7 cos
Environméntal issues Custorners
FIGURE 1. Strategic issues from an influence diagram perspective.
Figure 2 shows the interrelationships among the decision categories (the current
version of the model does not take into account either organisation or labour, which
are proposed as future areas of research).
Organis
ation i
\o Tecnology
Production planning and contro!
a
Vertical integration
FIGURE 2. Influence diagram reflecting the interrelationships among the
decision categories.
The general structure of the model is shown in Figure 3 (the detailed model is
available from the author). It shows, among others, some of the interrelationships
already mentioned.
As illustrated in Figure 3, the model is organised into six sectors:
1. Demand sector; computes the total demand based on type of customer, number
of customers, kWh per customer, and effect of price elasticity.
2. Generation sector; determines the mix of generated power (from hydroelectric
plants, thermoelectric plants, geothermal plants, wind generation plants and
private generation plants), taking into account losses in transmission and
distribution, available capacity, costs and load factors. It also dispatches the
available generation to meet load.
3. Installed capacity sector; keeps track of the installed capacity, by generation
source. It also considers depreciation and the reserve margin.
4. Cash flow sector; computes income, costs, debt, investment, interests and
interest payments.
5. Demand forecast sector; calculates the expected demand in the medium and
long term (i.e., five and ten years in advance, respectively).
6. New capacity addition sector; represents the construction and addition of new
capacity, taking into account demand forecasts, installed capacity, expansion
plans and the construction of capacity by private generators (who are not
supposed to provide, according to the new regulation, more than 30% of the total
demand).
The System Dynamics model was built using Powersim Constructor 2.51, and the
tuning and optimisation was achieved using its companion software Powersim
Solver 2.0.
GNP
Population
Diesel price
Demand Require- Generation
ments
* Residential LB + Priorities
* Industrial * Costs
* General Demand * Load factor
* Street lighting satisfaction * Losses
* Export L¢—__—___|
Rates leone Maximum
generation
Costs Capacity
Cash flow use
* Debt Installed capacity
+ Interests
* Income statement * Electricity source
* Amortisation * Capacity
* Reserve margin
+ Depreciation
Trends New A
Demand forecast
* Long term
* Medium term
Requirements
Capacity
Financing
construction
Available
Capacity
New capacity addition
* Expansion plan
* Maximum capacity by
source
FIGURE 3. General structure of the model.
MODEL USEFULNESS
The model had to "pass" the complete set of tests proposed by Forrester & Senge
(1996/1980), which deal with structure, behaviour and policy implications.
Furthermore, the set of summary statistics for evaluating the historical fit of System
Dynamics models proposed and explained by Sterman (1984) was used, as shown
below.
MODEL TUNING
The model assumes that parameters such as losses, margin reserve and load factor
are constant, whereas it can be observed that the values of these parameters
change within certain intervals. Figure 4 shows the combined evolution of the real
electricity generation (historical), that simulated by the model (simulated), and that
tuned by Powersim Solver 2.0. Figure 5 shows the combined evolution of the
installed capacity case.
6000
5000
4000
3000 —Historical
2000
1000 —+— Simulated
0
GWh
1968 1973 1978 1983 1988 1993 ——Tuned
Year
FIGURE 4. Historical, simulated and tuned generation data series.
2000
1500 —Historical
2 1000 —+— Simulated
500 Tuned
0
1968 1973 1978 1983 1988 1993
Year
FIGURE 5. Historical, simulated and tuned capacity data series.
Table 1 shows the numerical results obtained for the generation demand case.
Summary statistics are calculated for both the simulated and the tuned data series.
TABLE 1. Summary statistics: simulated and tuned electricity generation
Simulated Tuned
N 30 years 30 years
R? 0.98096 0.98370
Mean square error (MSE) 47820 32786
Root mean square error (RMSE) 219 181
Root mean square percent error (RMSPE) 0.07024 0.05804
Theil’s inequality coefficient (U) 0.07336 0.06075
Bias (U™) 0.22992 0.09438
Variation (U*) 0.09070 0.03718
Covariation (U‘) 0.67937 0.86843
Table 2 shows the summary statistics obtained for the installed capacity case.
Values are calculated for both the simulated and the tuned data series.
TABLE 2. Summary statistics: simulated and tuned installed capacity
Simulated Tuned
N 30 years 30 years
R? 0.93955 0.94130
Mean square error (MSE) 10727 13687
Root mean square error (RMSE) 104 117
Root mean square percent error (RMSPE) 0.13335 0.14206
Theil’s inequality coefficient (U) 0.14065 0.15887
Bias (U™) (fraction of MSE) 0.01809 0.06203
Variation (U*) (fraction of MSE) 0.23573 0.34349
Covariation (U‘) (fraction of MSE) 0.74617 0.59448
In both cases the numerical results allow the reliability of the model's behaviour to
be determined.
RESULTS
One of the purposes of the model is to generate insights in the analysis of several
policy implications. In this section some simulations are presented in order to
illustrate the nature of the results of the model.
Two cases are explored: (a) the impact of two different capacity expansion plans in
the total debt of the sector, one without more private generation investment, and the
other finding the optimum level of private generation investment, given certain legal
restrictions; and (b) the impact of a sensitivity analysis in the result obtained in (a).
Figure 6 shows the results of the first case, where the total debt of the electricity
monopoly might go from $725 millions in 1998 to $1069 millions in year 2020, if
private investment is suspended in 1998. On the other hand, if private generation
investment achieves a rate of approximately 30 MW/year, which is not contrary to
the new legislation requiring that total private generation be less than 30% of total
demand, and assuring full use of the private installed capacity, the total debt of the
Costa Rican government participation in the electricity power industry might go from
$725 millions in 1998 to $227 millions in year 2020. This optimum private
investment rate was found using the Optimize task of Powersim Solver 2.0.
1,200,000,000
1,000,000,000
800,000,000 a ae
® 600,000,000 i
400,000,000
200,000,000
0
1998 2002 2006 2010 2014 2018
Year
— Total debt without private investment
—- Total debt with optimum private investment
FIGURE 6. Comparison of the evolution of the total electricity debt under two
extreme scenarios.
Figure 7 shows how sensitive the "savings" obtained in the previous case are.
As explained before the model assumes that parameters such as losses, margin
reserve and load factor are constant along the simulation period, but actually the
values of these parameters change within certain intervals. The Assess Risk task of
Powersim Solver 2.0 is used, which allows for defining the assumptions of the
already mentioned parameters as statistical functions rather than specified values.
By using the historical ranges of these parameters, assuming triangular distributions
for all of them, it was possible to find the range that the total debt will fall between
with 100% certainty.
2,500,000,000
2,000,000,000
1,500,000,000
wn
1,000,000,000
500,000,000
0 +
1998 2002 2006 2010 2014 2018
Year
FIGURE 7. Range that the total debt will fall between with 100% certainty.
These results show that the Costa Rican government would be mistaken to assume
that private power generation will solve its debt problem. If the government does
not manage its electricity operations carefully the expected benefits of new
legislation might simply evaporate.
ELSECDYN: A MICROWORLD
A microworld (Morecroft, 1988; Senge, 1990) of the electricity sector of Costa Rica
was built using Powersim Constructor 2.51. The purpose of this microworld, called
ELSECDYN (an acronym for Electricity Sector Dynamics), is to aid different
stakeholders (i.e., legislators, managers, engineers, policy makers, among others)
to understand several interrelationships in the sector, and to give them the means to
experiment with the design of social and business policy in the safety of a simulated
environment.
Four of several control panels are shown in Figures 8, 9,10 y 11. Figure 8 shows
the main control panel, where decisions related to several parameters can be
changed before or during the simulation.
Load factor
(fraction)
Reserve margin
(fraction)
Losses
(fraction)
GNP growth
(%)
Population growth
(%)
Debt cost
(fraction)
Main control panel
04 05 06 07 08
a—i— dD
aver
a=
0.08 0.10 0.12 0.1
10
I=
I—i—d
0.06 0.08 0.10 0.12
FIGURE 8. Main control panel.
Figure 9 shows the control panel related to electricity costs classified by source of
generation (private generation is included as a special "source"), which allows for
different assumptions about these costs to be tested.
Figure 10 shows the demand multipliers, which allow experimenting with different
than expected growth rates in residential, industrial and general electricity demand
sectors.
10
Electricity generation costs
($/GWh)
Private @ —_|— b> 6.404
Wind X | —__— b> 4.5e4
Geothermal @ —}—— > 2e4
Hydro ah D Itee
Thermal @ —_—_ > 9.9e4
FIGURE 9. Electricity generation costs panel.
Local demand: multipliers by sector
Local
demand Residential Industrial General
mail G19)» (GD) G9)
0
FIGURE 10. Demand multipliers.
Figure 11 shows the control panel dealing with export demand and private power
generation demand. This panel allows the users to see, for example, the impact in
the Costa Rican electricity power sector, of exporting electricity to other Central
American countries.
11
Export demand Private generation capacity
400 200 400
FIGURE 11. Export demand and private generation demand panel control.
ELSECDYN allows the effects of changes in many parameters in several parts of
the electricity sector to be seen. Some of these parts can be as diverse as total
generation costs, accumulated debt of the government participation in the electricity
sector, annual interests charge, total generation, installed capacity and unmet
demand. Figure 12 shows an example of the simulated results of the electricity
sector debt balance, in both graphic and table form.
Debt balance
1,000,000,000.00-
‘* 500,000,000.0
0.00.
2,000 2,005 2,010 2,015 2,020
Years
Time DEBT
2,016 $994,169,585.49 I
2,017 $976,406,995.44
2,018 $997,115,816.64
2,019 $1,056,610,153.31
2,020 $1,096,940,503.40 >
ft LI
FIGURE 12. Electricity sector debt balance simulated results.
12
CONCLUSIONS
The System Dynamics model of the Costa Rican electricity power sector, allows,
following structural and behavioural validation, the simulation of several policies
intended to improve the overall management of this national sector.
It was shown how simply launching new legislation will not provide sustainable
solutions to the electricity debt problem of Costa Rica, because the interaction of
other variables might well counteract the expected results of the new regulation.
It was also found that there is an unquestionable relationship between strategic and
operational issues. Typically, only one of such approaches (strategic or operational)
is pursued. However, this paper shows how the strategic issues are actually
achieved when confronted with day to day operations.
It is normal to define common sense strategic paths, but the experimentation of how
such Strategic options can develop along time can be enhanced if a System
Dynamics model is defined, validated and used as a testbed of the different
strategic options.
Finally, it was shown how the use of tools such as tuning, optimisation, and risk
assessment can help in designing better models for policy making.
REFERENCES
Aslam, J. y Saeed, K. (1995). Electricity conservation in domestic sector of
Pakistan: A System Dynamics approach. Proceedings of the International
System Dynamics Conference (Vol. 2, pp. 333-341). Tokyo, J apan.
Barton, P. y Bull, M. (1986). Planning conservation programs decision
support with the Conservation Policy Analysis Model. Proceedings of the
International Conference of the System Dynamics Society (pp. 661-669).
Seville, Spain.
Bunn, D., Larsen, E. y Vlahos, K. (1997). Complementary modelling
approaches for analysing several effects of privatization on electricity
investment. In D. Bunny E. Larsen (Eds.), Systems modelling for energy
policy (pp. 303-325). Chichester: J ohn Wiley & Sons.
13
Coyle, R. (1996). System Dynamics modelling: A practical approach.
London: Chapman & Hall.
Dyner, I. y Bunn, D. (1997). A systems simulation platform to support energy
policy in Colombia. In D. Bunny E. Larsen (Eds.), Systems modelling for
energy policy (pp. 259-271). Chichester: J ohn Wiley & Sons.
Ford, A. (1990a). A method for approximating the value of demand-side
information. Energy Systems and Policy, 13, 135-151.
Ford, A. (1990b). Estimating the impact of efficiency standards on the
uncertainty of the Northwest electric system. Operations Research, 38, 580-
597.
Ford, A. (1997a). System Dynamics and the electricity power industry.
System Dynamics Review, 13(11), 57-85.
Ford, A. (1997b). The changing role of simulation models: The case of the
Pacific Northwest electric system. In D. Bunny E. Larsen (Eds.), Systems.
modelling for energy policy (pp. 241-258). Chichester: J ohn Wiley & Sons.
Ford, A. y Bull, M. (1989). Using System Dynamics for conservation policy
analysis in the Pacific Northwest. System Dynamics Review, 5, 1-16.
Ford, A., Bull, M. y Naill, R. (1987). Bonneville’s conservation policy
analysis models. Energy Policy, April, 109-124.
Forrester, J. y Senge, P. (1996). Tests for building confidence in System
Dynamics models. In G. Richardson (Ed.), Modelling for
management: Simulation in support of systems thinking (Vol. 2, pp.
413-432). Cambridge: Cambridge University Press. (Original work
published en 1980).
Grupo de Dinamica de Sistemas de la Universidad de Sevilla. (1993).
Memoria del proyecto MEREA. Documento interno.
Larsen, E. & Bunn, D. (1999). Deregulation in electricity: understanding strategic
and regulatory risk. | ournal of the Operational Research Society, 50(4),
337-344.
Lyneis, ]. (1997). Preparing for a competitive environment: Developing strategies
for America's electric utilities. In D. Bunn y E. Larsen (Eds.), Systems
modelling for energy policy (pp. 273-301). Chichester: J ohn Wiley & Sons.
Morecroft, J. (1988). System Dynamics and microworlds for policy makers.
European | ournal of Operational Research, 35, 306-308.
14
Naill, R. (1992). A System Dynamics model for national energy policy planning.
System Dynamics Review, 8, 1-19.
Pérez Rios,]. (1999). Modelado integrado de sistemas: Combinacién de
metodologias cualitativas y cuantitativas para el estudio de sistemas
dinamicos. Ill] ornadas de Ingenieria de Organizacion (Vol. 1, pp. 125-132).
Barcelona, Espana.
Senge, P. (1990). The fifth discipline: The art and practice of the learning
organization. New York: Currency Doubleday.
Sterman, J. (1984). Appropriate summary statistics for evaluating the
historical fit of System Dynamics models. Dynamica, 10, 51-66.
Wheelwright, S. (1984). Manufacturing strategy: Defining the missing link.
Strategic Management] ournal, 5, 77-91.
Wheelwright, S. y Hayes, R. (1985). Competing through manufacturing.
Harvard Business Review, 1, 99-109.
15
