Load Frequency Control in Deregulated Power Systems
Load Frequency Control in Deregulated Power Systems
Abstract-Open transmission access is a legal requirement in PURPA is the emergence of independent power producers
the United States, but is not fully implemented. Discussion of (IPPs) that can sell power to W s and these are also shown
deregulation has so far focused principally on the tariff struc- in the figure. Thus, all the square boxes in the figure denote
ture for transmission access, but operating the power system in business entities which can buy andor sell electric power.
this new environment will present significant problems of an
Finally, it shculd be mentioned that VIUs can come in many
almost purely technical nature. Something as simple as fre-
quency control becomes challenging when implemented in the sizes but the thousands of co-operative and municipal utili-
competitive, distributed control environment that true third ties in this broad-brush picture can be considered, just like
party wheeling creates. This paper seeks to identify likely de- large industrial sites, as customers that may also have IPP
regulation scenarios, identify the technical issues associated type generation.
with Load Frequency Control, and identify technical solutions, Vertically Integrated Utility IPP
such as standards and algorithms, needed for the operation of
this key component of national infrastructure in the face of
profound structural changes.
I. INTRODUCTION Tie Lines Transmission Tie Lines
pJEpJ
largely in the hands of vertically integrated utilities (VIUs) Transco
which own generation-transmission-distributionsystems that
supply power to the customer at regulated rates. Such a con-
figuration is shown conceptually in Figure 1, in which the
large rectangular box denotes a VIU. The VIU is usually in- Genco
terconnected to other W s and this interconnection is al- Genco
most always at the transmission voltage denoted in the figure Figure 2 - Deregulated Utility Structure
as tie lines. Thus, electric power can be bought and sold be- Given the present situation, it is generally agreed that the
tween VIUs along these tie lines and moreover, such inter- first step in deregulation will be to separate the generation of
connection provides greater reliability. power from the transmission and distribution, thus putting
The major change that has happened since the passing of all the generation on the same footing as the IPPs. We have
designated these in our deregulated picture of Figure 2 as
Gencos which will compete in a free market to sell electricity
they produce. The rest of the scenario is less clear and may
take some tiime to evolve. It can be assumed that the retail
customer (e.g. residential) will continue for some time to buy
from the local distribution company. Such distribution com-
panies have been designated Discos. The entities that will
wheel this power between gencos and discos have been des-
ignated Trunscos. Although it is conceptually clean to have
This paper was presented at the 1995 IEEE Power Industry
separate functionalities for the gencos, transcos and discos,
Computer Applications Conference held in Salt Lake City, Utah, in reality there will exist companies with combined or partial
May 7-12,1995. responsibilities. This does not take away from the generality
of this model which we use in this paper to discuss technical A variable power contract is defined here as one in which
issues. For example, the difference in wholesale and retail the customer can ask to vary the power it buys within certain
wheeling in this scenario is only in the number and size of parameters specified in the contract. This would allow the
transactions and not in the technical issues (although the customer to match the purchase with the power demand.
increased magnitude of computerized accounting needed by The third type is called here the load matching contract
the explosion in the number of transactions is, in itself, an- and can be understood in the context of load following. Since
other type of technical issue). In fact, this generality is not load following today is a VIU responsibility not specifically
lost if we only consider all bilateral contracts between discos paid for by customers, it is difficult to see how this will be
and gencos, as direct contracts between customers and gen- handled in a deregulated environment. One possibility is for
cos are the same from the technical sense. the customer to be responsible for one's own load following
In Figure 2 only one transco is shown but in North Amer- in which case the contract with the genco would require the
ica the transcos will continue to be interconnected and, in genco to follow the load by closely matching its generation to
general, transactions will flow through more than one it.
transco. This is somewhat different from the UK or Norway 11. REALPOWER
CONTROL
model where the breakup of a government owned system
made it possible to have one transco. The technical advan- Any power system has the fundamental control problem of
tage of having one transco is significant as the responsibility matching real power generation to load plus losses, a prob-
and authority for system performance can then be completely lem called Load Frequency Control (LFC) or frequency
assigned to one entity. In England, the system has gone even regulation. To avoid confusion between frequency regulation
further in making the transco (actually, the National Grid and power industry deregulation, the abbreviation LFC will
Co.) the sole broker for all transactions. However, it can be used. Cohn [l] gives control algorithms for this problem
safely be assumed that the scenario in the US will not allow in the VIU which form the basis of present LFC. Cohn's
this simplification and a many transco general system is as- structure assumes that the control area, defined by the physi-
sumed in this paper with all possible bilateral contracts be- cal transmission facilities owned by a utility, has the author-
tween gencos and discos, ity to issue real power control signals to all of the generators
Finally, we should note that this paper is not about how to inside itself, which are also owned by the utility. This will
price either energy or any other service. This is a major topic not be true in deregulated scenarios.
that is getting much attention already but this paper is about A . Possible Structures
the technical issues of operating the power system under
There are many possible structures for providing real
such a deregulated scenario. We assume that some pricing
power control in a deregulated environment, differentiated
mechanism can be established either by free market, by vol- by who controls generators, who has the obligation to per-
untary agreements, or by specific government regulation. form LFC, and how free the market is. We consider three
These are assumed to be part of all the bilateral contracts we examples, 'Ifree" LFC, '%barged" LFC and bilateral LFC,
consider. The thrust of this paper is to identify technical
progressing from no LFC market to a fully distributed mar-
issues that have not had much exposure in the literature and
ket, that cover the range of possible LFC structures. Free
to suggest some possible ways to handle them under these
LFC is the present situation, where a VIU (or transco) issues
new circumstances. The following discussion identifies some
real power control signals to units it owns to perform LFC,
general types of bilateral contracts that are needed in raising
with no specific cash flow tied to this function. IPPs in this
the technical issues. environment are either captive to the VIU and contractually
A . Technical Contract Types obligated to accept control as though they were VIU-owned
In practice, there will be an infinite variety of contracts. generators, or have no obligation to perform LFC. There are
The variety in today's regulated conditions is quite large; the some obvious economic inequities in this arrangement, but
types in a deregulated environment can be barely imagined. from a technical perspective, LFC will not be a significant
However, for this technical discussion only three types of problem until the proportion of VIU controlled units in the
contracts need be considered. At present, the retail customer generating mix decreases to the point where the total con-
pays for energy at a fixed rate without any concern for how trolled generator response rate drops to near the maximum
the power demand varies over time. Wholesale customers load change rate.
may have to pay for power factor, maximum demand and "Charged" LFC is a structure in which the transco no
other characteristics. longer owns generation, but has an obligation to provide
The simplest contract is for $xed power over a fixed time LFC. In this case, the transco must purchase real power from
period. Since very few customers consume power at a fixed gencos, and resell it to the discos, on a short term, real time
rate, it is obvious that the customer will need more than this. basis (seconds). This structure can operate with a transco
monopoly on the purchase and resale of LFC power, with or
1193
without bilateral contracts, as is the case in Norway and the before, while (because the area is net importing) the gencos
United Kingdom, respectively. [2,3] In bilateral LFC, the are likely to sell locally. For net power exporting areas, the
transco has no obligation to provide LFC. Discos must pur- discos will mostly buy internally, while the gencos will sell
chase load matching (LM) contracts from gencos. It is pos- both locally and outside the boundary. Thus the number of
sible for such bilateral LM contracts to co-exist with the boundary crossing transactions will rise by about the number
transco obligation. In this case the transco is obligated to of discos or gmcos connected to a transco, which is likely to
provide LFC to discos that have not purchased LM contracts. be in the tens.
B. "Charged"LFC The existing methods of processing transactions will re-
quire review to ensure that predicted transaction rates can be
In "charged" LFC, the transco executes the control algo- accommodated. In particular, there appears to be a strong
rithm, measuring frequency deviation and net interchange need to enter transaction data once and communicate it
deviation, and then issuing control signals to gencos to ad- electronically to affected control centers in order to reduce
just their output. Calculation of control error is identical to the entry error rate.
the "free" LFC case, because the control area boundary is the The second issue in "charged" LFC occurs in the alloca-
same, enclosing the transco and gencos and discos attached tion of controll error to generators. In "free" LFC the control
to it (Figure 1). Control error is given by: error is allocated based on economic dispatch, that is, the
generators arc: moved which provide the necessary control at
where ACE is area control error, positive values requiring an the lowest fuel cost. In "charged" LFC the same basic prin-
increase in internal generation, NI is net interchange, B is ciple will apply, that is, the generators will be moved which
bias (in M W / O . l Hz, having a negative value) and f is fie- provide the necessary control at the lowest cost, but this will
quency. be the ~ ~ ~ ~ S Ccost
I D 'toS purchase LFC energy from the gen-
cos, set by some form of bidding or market. Although the
Gencos @ allocation problem is mathematically simpler than the "free"
Control Boundary
-
!2 Transoo
1 1 1 :lis 1 :yk
fect at any given time at a VIU. The number of transactions, boundary is zero. Consider the case where DA and DB loads
and the rate of new transactions already causes concern over are exactly 100 MW. Total generation will exceed this due to
the accuracy of the transaction data, presently entered by losses, and control error will cause G1 to rise until it is sup-
hand at both of the control centers party to the transaction. plying the losses:
When erroneous values are entered, the error can have a G r Gmnitig fSet;$
large effect after a long time period. [4]
Even in the short term deregulated environment, it seems 1ooMw 1ooMW
likely that the number of boundary-crossing transactions will Losses 2MW
increase to hundreds or thousands. For net power importing Total 202MW 202MW 202 Mw
control areas, each disco is likely to have about as many In this cointrol cycle, the transco has purchased 2 MW of
boundary crossing contracts as the entire control area had LFC from Cil, and then sold 2 MW of LFC to itself, since
1194
the transco is responsible for its own losses. The transco MW to itself. G1 is paying the cost for not responding. The
could have contracted for power to provide its losses, instead LFC sale to G1 is
of relying entirely on LFC power. Therefore the LFC loss LFC Sale,, = Setpoint,, - Generation,, (6)
sale is
Loss Sale = cGeneration - Load
(31
Accurate billing will require accounting software to track
the LFC purchases and sales of each participant, genco,
-xJnterchange - 1Loss Contracts
disco and transco, in terms of energy purchased and pur-
chase prices on a control cycle by control cycle basis. This
The LFC purchase from genco G1 is software will need to h o w the actual values of generation
LFC Purchase, = Setpoint, - IContracts, (4) and load for each control cycle, the participants and contract
power values for all contracts, and the LFC pricing mecha-
Suppose that DA load increases by 2 MW. Control will nism.
cause G1 to rise to 104 MW. The actual generation values are already telemetered to
Genco I Generation I Setpoint I Disco 1 Load
control centers at the LFC cycle rate, usually every 2 or 4
G1 I 104MW I 104MW I D A I 102MW
62 1OOMW 1OOMW DB 1OOMW
seconds, as are the control area tie lines. Disco loads are not
Losses 2MW presently telemetered, or are telemetered at slower rates,
Total 204MW 204MW 204 MW usually every 12 seconds. Assuming that the number of dis-
cos is about the same as the number of gencos or tie lines, a
50% increase in fast telemetry will be required for accurate
accounting. Telemetry from gencos and discos outside the
control boundary is not needed, although the contract power
LFC Sale,, = Load,, - XContracts, value for the side of the contract (disco or genco) inside the
It is possible for the load increases and decreases at different control boundary is needed at the control center.
discos to sum to zero. The transco is then selling positive It is likely that discos will not be content to purchase LFC
LFC to some discos and negative LFC to others, even with from one source, the transco. Instead they will want to make
no control error. load matching (LM) contracts with gencos. The disco must
When DA's load first increases, governor action will cause then send its required generation value to the genco every
GI and G2 to increase in power before LFC raises G1. If control cycle. That value must also be known at the transco
each generator has identical governor response, values prior control center every control cycle, since it is the contract
to control action will be power value for the LM contract for the control cycle. A
Genco Generation Setpoint Disco Load tradeoff between computation and communications is possi-
G1 103MW 102MW DA 102MW ble if the contract value is a simple function of the disco
62 101MW 1OOMW DB 1OOMW load, which is already telemetered to the control center.
I I Losses I 2MW When an LM contract crosses a control boundary, a transac-
tion value every control cycle will be needed by the genco's
The transco is purchasing 3 MW of LFC from G1 and 1 transco.
M w from G2, even though there is no LFC offer made by
C. Bilateral LFC
G2. Thus every genco in the system has an implicit LFC of-
fer with the transco! Clearly the price for such involuntary In the bilateral LFC structure, the central issue is that
LFC power purchase should not be set by the gencos, or they there is no central control algorithm. Each disco must pur-
can set it to an unreasonably high figure, nor can it be set by chase LFC from one or more gencos. Control is highly de-
the transco, or it would not pay at all. This will be a regula- centralized. Each LM contract requires a separate control
tory pricing issue. process, yet these control processes must cooperatively
Suppose that the transco issues control to 61, and it fails interact to maintain system frequency and minimize time
to respond: error. An example of cooperative decentralized LFC already
Genco Generation Set oint Disco exists in the interconnected system of W s , which solves
essentially the same problem, albeit with fewer participants.
101MW 10oMW In this system, a separate control process exists for each
Losses 2 MW control area. Since a separate control process is needed for
Total 204MW 204MW 204 MW each LM contract, there must be a control area associated
In the previous case, the transco bore the excess LFC cost with each contract.
because it had not yet issued a control signal. Now, the These control areas are virtual. They come into existence
transco is purchasing 4 M W of LFC from G1, and 1 MW at the start of an LM contract, and terminate at its end. The
from G2, and it is selling 1 MW to G1, 2 MW to DA and 2 boundary of the virtual control area (VCA) encloses the
1195
genco and the disco associated with the contract, but the "tie Inadvertent interchange will be as important an issue as it
line flows" across the boundary are not the measured flow is today. Mismatches between actual generation and genera-
values, but rather that portion of them assigned to the con- tion setpoints appear as inadvertent interchange, as is the
tract. case in VIU LHC systems. For example, a 5 MW increase in
102yloop y q-q@I Gencos
DA load would initially be met by the governor responses of
all generating units in the interconnected system. Assume
that G1 and G2 each pick up 1 MW of the load increase.
Then GO (representing all other generators) picks up 3 M W
(Figure 5 ) .
Physical Virtual
Figure 4 - Virtual Control Areas in Bilateral LFC
Figure 4 illustrates VCA boundaries for the following
contract scenario:
G1 supplies 60% of the LFC requirement of DA
0 G2 supplies 40% of the LFC requirement of DA
Physical Virtual
0 G2 supplies 100% of the LFC requirement of DB
-
Figure 5 Load increase, governor response
0 G1 has a 90 MW fixed power contract with DB Here, the (G1-DA transaction has -2 MW of inadvertent
0 G2 has an 80 MW fixed power contract with DA interchange, and the G2-DA transaction has -1 MW. The
G2 supplies 100% of the LFC requirement of the distribution of G2 output among transactions is assumed to
transco, for transmission losses recognize the: change in DA and allocate all the inadvertent
DA and DB have loads of 100 MW each to that transaction, instead of distributing it among all trans-
GI and G2 outputs are at their control setpoints actions involving G2. All other generators together have 3
No tie line flow is occurring MW of inadvertent interchange. Interchange schedule errors
Figure 4 shows the physical power flow values for the dis- and metering: errors at the discos can also result in accumu-
cos and gencos, and VCA interchange values. The solid lation of inadivertent interchange.
VCA boundaries denote LM contracts, while the dashed A key difference between the bilateral LFC environment
boundaries are fixed or variable power contracts. and the VIU environment is that the control areas in the lat-
The fixed or variable power contracts operate on the as- ter are persistent, while the control areas in the former last
sumption that the contracted power is generated and con- only while the contract is in operation. This requires that
sumed. No control is required, and no communications dur- bilateral LN[ contracts pay back their inadvertent inter-
ing the contract if the power value can be predicted at each change in kind before the contract is terminated. This pre-
control cycle. The contract value affects the division of sent two problems. First, schedule and metering errors are
physical flow among the VCA interchange values. usually detected well after the fact, for example by compar-
The control algorithm for each VCA is executed at the ing MW-hour metering with integrated MW metering. The
genco end, since ultimately the genco must adjust the gover- control area responsible for the resulting inadvertent inter-
nor setpoint(s) of its generator(s) for LFC. The disco end of change may no longer exist - even virtually. Second, VIU
the VCA requires a moderately intelligent meter that takes control areas, attempt to pay back inadvertcnt interchange at
the measured power value and subtracts all fixed and vari- similar time periods, with the intent of paying back energy in
able power contract values to obtain its total LM requirement similar cost regimes. A bilateral LFC control area may not
(which may be negative). The meter then divides the LM re- exist long enough to reach the next similar pricing period.
quirement among the LM contracts, and sends contract LM A solutiori to this problem is to accumulate inadvertent
requirement values to the genco on the other end of each interchange for each genco, rather than for each transaction.
contract. There are several justifications for this approach. The gencos
Transco loss calculations are a minor issue. They can be are persistent, and can pay back metering and schedule er-
calculated by summing generation minus loads and inter- rors after they are detected, passing the cost of payback along
change, as in the "charged" LFC case. It is perhaps more to the responsible entity, if necessary. Gencos are responsible
likely that transcos will calculate losses from state estimator for inadvertent interchange due to insufficient generator
voltage solutions, using loss coefficients to update loss values response, because in an LM contract they assume the re-
between new solutions, trading more computation for less sponsibility of matching the disco's load. Therefore there is
communications. no reason to involve the disco in control error related issues.
1196
contrast to the casual aside by the authors in the f i a l The canmdqu&ions above are intended as thought-provokers' to suggest that within the
paragraph of their Section I, the disco would be responsible new business environmentthere are some physical realities that need to be taken care of, but
at the same time, how the implementation occurs may be quite different fiwnwhat we are
for i t s own load regulation (not load following, which deals currentlyaccustMnedto. I would be interested in the authorsthoughts on the points raised,
with the diurnal load pattern), but not by imposing and wishthem well in further work they may do in this area
impossible demands on their supplying genco.
Manuscript received May 26, 1995.
This pattern of behavior would have the added advantage of
minimizing flows of regulating energy on the bulk power Gerald B. Sheble' (Iowa State
system. There will, of course, always be residual flows o f University, Ames, Iowa) The authors
regulating energy. We suggest that, as an inescapable have written a very timely paper
concomitant of system operation, the transco will contract for dealing with a central concern of
sufficient responsive generation to satisfy these needs, and deregulation. However, the authors
embed the cost in its tariff. (The tariff could also include
have not included several papers
which dealing with control of
penalties for excessive use of system regulation.) This
jointly owned units, the referenced
practice would complement in a natural m e r transco paper contains excellent discussion
contracts (which could be viewed as de facts "insurance of such control schemes [ll. It
policies"') for spinning reserves necessary to maintain system also seems similar to many control
security. adaptations to cover the non-native
load component included within some
Such arrangements would seem to offer attractive options that control areas as depicted in Figure
would reflect inbuilt incentives in a restructured industry, 1. This figure shows two control
would simplify both the control and communication areas with non-native demand in area
requirements of the authors' second, "charged" LFC, 1 that is served by generation in
scenario, and could preclude, or at least minimize any area 2. Such pockets of demand are
perceived incentives for, consideration of their third, created as companies merge. The
bilateral LFC, scenario. company responsible for the
non-native demand in located in area
References 2. The solution is to add a
"dynamic interchange contract" for
D1. [Link]&: Impact of Electric Utility Restructuring the non-native demand such that any
on Energy Management and Generation Control; demand changes are picked-up by
Engineering and Operations Workshop, APPA generation in area 2 and not in area
Communications & Control Comm., Philadelphia, 1. I would like for the authors to
Pennsylvania, 6 April 1995. comment on the similarities of these
command adaptations. I believe that
D2. [Link], [Link]: Synchronous Generator they are extremely similar to those
Capability Curve Testing and Evaluation; PEEE presented within the paper.
Trans. [Link]-9 n. 1, pp.414-24.
units that trip due to faults. How will reservesbe provided, engaged and p r i d 7 "Automatic Generation Control of
2) What role could interruptible loads play in the future of LFC 7
3) &dI: energy contracts be written, both by suppliersand cotlsumers of power ( g e n m Jointly-Owned Units,'' IEEE
and discos) as a base amount with the control requirementsCONactedspecificallyby Transactions on Power apparatus and
robust, quick respondingunitsdesigned for the purpose ? Metering and accounting
would be simplified if base units were set not to respond to load and fiqency changes,
Systems, Vol PWS-98, No. 1, Jan/Feb
leaving those fbctiom totallyto the "control units". Discos meter and account for the 1979, pp 207-218.
base energythey contractedfor, plus the "ripple" or control energy, used for load
deviationsfiom the base. There are different prices for "base energy" and "control Manuscript received June 9, 1995.
energy". There are no "inadverlents", in the traditional sense, only control energy, bought
and sold.
4) The entire interconnectionoperates as one large control area, instead of the multitude
of control areas per NERC region as we now have. Designated units provide frequency Shangyou Hao, Alex Papalexopoulos and Harry Singh (PaciJic
control, as h i d above, while LFC and interruptibleloads become a tool for
transmissionsecurity implementation, to ensure that voltage and line flows are kept Gas and Electric Company, San Francisco, CA): The authors are
secure. to be commended for attempting to address an important technical
1199
issue that is at the center of the debate of developing an ancillary possible contracts. We wished to decouple the technical as-
services market for facilitating transmission open access. We pects of a ccintract from its financial aspects. We note the
would appreciate the authors’ comments on the following correspondence of fixed and variable contracts to economy
observations: energy for base load and diurnal variations, and load match-
ing contracts, to regulating energy. (We avoided the term
1) The authors consider the supply of losses to be a part of Load
“regulation” in the paper so as not to confuse it with rate
Frequency Control (LFC). This may be true for incremental losses
only. In general, the supply of losses does not have to be part of the regulation, and instead used the terms “load following” or
LFC ancillary service. As the industry further unbundles its “load matching,” perhaps introducing a different confusion.)
products and services, transmission loss compensation will be Mr. Fink iraises the issue of incentives for the genco to
managed as a separate service. provide control. Of course, the best incentive is money. As
Messrs. Hao, Papalexopoulous and Singh point out, providing
2) In the discussion of charged LFC, the allocation of control error control is expensive. This leads Mr. Fink to conclude that
to generators assumes participation from all generators. This gencos would prefer fixed or, at the most, variable contracts.
assumption is unrealistic since it is uneconomical for all generators This means that discos and other customers will have to be
in a control area to be controlled by LFC. In order to implement a prepared to pay the price that will provide enough incentive
charged LFC structure, the price for LFC services should be for the suppliers of control. Transcos (or poolcos) may have to
carehlly evaluated to reflect all incurred costs by generators. coordinate the control and its marketing, as in the charged
These costs should include communication and hardware costs, a LFC scenario, but the price is always paid by the customer.
portion of generation plant capacity and operating costs, and
possibly opportunity costs of the plant. However, the associated AlternativeIy, the requirements for control can always be
energy costs should be recovered separately, possibly from the mandated to the gencos, but this would be against the spirit of
energy deviation accounting process. an open marketplace.
-
A general itheme in our paper is that the loads the discos -
3) The bilateral LFC structure proposed in the paper assumes no are responsible for purchasing the services they require. This
central control within a control area. Requiring from each “Load point bears on several of the discussions. We are in agreement
Matching” contract a separate control process may be appealing with Mr. Fir& that discos must be responsible for their own
but it is not realistic. In any case, under any such scenario we still load regulation. Mr. Fink’s description of discos finding eco-
have the need to assign the overall obligation of meeting the nomic benefit in supplying most of their own load matching
“system’s AGC responsibility to the interconnection” to one entity seems quite possible. Moreover, just as the disco is responsi-
that will act as the “back stop” service provider. That entity will ble for load matching, we expect that it will also be responsi-
probably be the local grid company that will have auditing rights
ble for its own spinning reserve requirements, (an issue also
to make sure that AGC providers (the mini control areas) meet
their obligations. Rules and guidelines are then needed to provide raised by Mr. Jamniczky) and may well prefer to supply these
the proper incentives so that these AGC providers are good locally, or purchase them from a genco, rather than being
performers and that undue reliance on the grid company is avoided. constrained 1.0 purchase them from one supplier.
These rules should very clearly incent the AGC providers to meet The responsibility to purchase required services also bears
the operational standards of frequency control. Otherwise, system on the issue of losses raised by Messrs. Ha0 et al. In our ex-
reliability will be jeopardized. amples, we consider the losses to be part of the implicit pur-
chase agreement that the transco has with the gencos of the
Manuscript received June 23, 1995 system. This would correspond to the incremental losses
mentioned in the discussion. We also state that the transco
Richard D. Christie (University of Washington) and An- could have an explicit contract for energy purchase to cover
jan Bose (Washington State University): We thank all of the losses. This contract could be for some or all of its losses.
discussors for their input. We are happy to see that our efforts This would correspond to the separate service in the discus-
have had the effect of energizing what we feel is an important sion. We recognize that there may be different ways in which
dialog on the technical concerns associated with structural the losses are supplied and accounted for without any direct
reorganization of the utility industry. Our closure should be effect on LFC.
considered, not the final word, but perhaps the next statement We thank Professor Sheble’ for pointing out the similarities
in a continuing discussion. between the control schemes we discuss and previously exist-
We appreciate the humor, as well as the diffculty, involved ing Jointly Owned Unit (JOU) and non-native demand control
in holding a satisfymg discussion of LFC without satisfying problems. Tlo the extent that bilateral contracts require one or
system constraints. Mr. Fink is correct that we consider only more physical generators to accumulate demand from physi-
free-flowing ties. We felt it was a reasonable assumption that cally non-contiguous areas to determine operating setpoints,
a contract for the sale of energy from a supplier to a consumer there is much to be learned about algorithms and communi-
would have to satisfy security criteria (hence network con- cations from this prior work, which has dealt with modifymg
straints) prior to its implementation. We note in passing that existing control schemes based on persistent control areas
most LFC papers ever written also make this simplifying as- with physical boundaries to accommodate special cases. The
sumption. differences in the control schemes we address are perhaps
The basic contract types we discuss, fixed, variable, and found in a basic concept - the Virtual Control Area (VCA) -
load matching, are also simplifications of the many types of which can have a limited existence and time-varying, concep-
1200
tual boundaries (only a portion of a load, for example, may be this is not an insurmountable obstacle, and we have overcome
in a VCA, and that portion can vary with time). The VCA it with the VCA concept. There is no llsystem responsibility to
concept allows us to apply established LFC concepts and al- the interconnection" belonging to an entity superior to exist-
gorithms even to a contract-driven, full bilateral LFC control ing control areas now, and there is no "back stop" service
scheme. We believe that the JOU and non-native demand provider in the western U.S. (or elsewhere), yet system reli-
control schemes could be easily expressed in terms of VCAs, ability is adequate. There is no a priori reason to believe that
while the persistence of the control areas in prior work would there would be a need for these functions in a full bilateral
not be suitable for dealing with the deregulated environment. LFC structure, or that the reliability of the power system will
The participation of all generators described in the charged be less simply because there are many more control areas. We
LFC scenario is based on the natural governor response of the do believe, however, that the information exchange (of meas-
generators, or speed droop, not on control action. Control is urements and control signals) needed for full bilateral LFC is
issued only to generators from which energy is purchased to very high and further study of costs versus benefits is needed.
meet control demands. A flat droop characteristic, if that is Messrs. Ha0 et al. raise the important issue of establishing
what is being suggested by Mr. Jamniczky, is not advisable and enforcing requirements for good control performance.
for synchronous generators. Therefore all generators will de- These monitoring functions are presently performed by the
viate from their setpoints in response to load changes, hence reliability councils, such as WSCC. If this policing function
the implicit LFC contract described in the text. We suggest at present is not particularly strict, it is only because the
that the inadvertent energy accumulate to the gencos, and be member companies have wanted it that way. We see no rea-
repaid in kind, following present practice. However, payment son why a similar entity (poolco or independent system opera-
in cash' is also possible, and incentives for better control can tor) cannot continue to perform this task in the future, al-
be built into a cash system in the form of fines and rewards. though the magnitude of the task will grow and require some
We note the reluctance of some of the discussors to con- adaptation. Given the existing difficulties in assigning re-
sider the possibility of full bilateral LFC. Mr. Fink hopes to sponsibility for bad control performance to individual control
preclude consideration of it, and Messrs. Hao, ef al. term it areas, we believe that the problem of accurately detecting
"not realistic." We would like to make it clear that we are poor control performance in an environment of competing
neither advocating nor opposing full bilateral LFC, but there entities presents significant technical concerns, and hence
appears to be no technical reason that precludes a full bilat- analytical challenges. We believe that this point also requires
eral LFC scenario, and a number of fundamental economic more study.
reasons to expect that such a structure will evolve. Full bilat- We thank our discussors again for their thought-provoking
eral LFC is theoretically identical to the present LFC struc- inputs, and look forward to continuing this discussion as the
ture, which consists of numerous control areas operating in- course of deregulation moves forward.
dependently on one synchronous physical system, such as the
western United States. While the present control scheme as-
sociates fixed geographic boundaries with these control areas, Manuscript received November 30, 1995.