Nof S.Y. Springer Handbook of Automation

Подождите немного. Документ загружается.

Securing Electrical Power System Operation 65.2 Ancillary Services Planning 1155

65.2.2 Minimal Needs of AS

In order to determine the power reserves in the form of

AS that would technically sufﬁce for proper power bal-

ance control with acceptable reliability, it is necessary

to refer to reliability standards. Since the UCTE has no

speciﬁc reliability standards for its members [65.18],

seven indices describing reliability are deﬁned in Ta-

ble 65.3 for this purpose. Each of them represents

a probability that ACE exceeds a certain threshold and

lasts for a certain amount of time (value at risk). The

index rACE

refers to the one-minute average value of

ACE, while theindexrACE

refers to an average value

of ACE over 1h.

The indices rACE

and rACE

are statistical mea-

sures of ACE that are related to NERC’s CPS2. rACE

is linked to the UCTE requirement for frequency recov-

ery in less than 15min after a forced generation unit

outage, and thus it is similar to NERC’s DCS index. The

other indices (rACE

60 tx

) indicatethe possibledanger of

a longer-lasting powerimbalance that maylead toan ap-

plication of load-shedding mechanisms or to a blackout

in the worst case.

–

Σ min

Not allowed Not allowed

(200 MW, 87 %)

100

Reserve Reserve

(%)

1Cumulative distribution function for ACE

Acceptable range of ACE irrespective of number of cases

ACE

(MW)

Acceptable number of cases

when ACE

> limit

100–rACE

rACE

Acceptable number of cases

when ACE

≤ limit

Cumulative distribution function for ACE2

Area control error < 200 MW found in 87% of all observations (an example)

Fig. 65.9 Cumulative distribution function of the open-loop ACE and its use

The desired values were determined statistically

from historical recordsof ACE. A few consecutive years

should be evaluated.The durationof theperiod assigned

for statistical evaluation should correspond to a period

when the technical parameters of the transmission sys-

tem and the market conditions are compatible with the

current situation and practices. In addition, no prob-

lems related to the area performance should be reported

from either TSOs in adjacent areas or from domestic

customers. Hence, we can assume that the TSO’s per-

formance was satisfactory.

The indices are direct measures of the quality of

power balance control in the control area, and indi-

rect measures of the reliability of the power supply to

end-consumers.

The AS categories considered throughout the chap-

ter are described in Table 65.2. Analytical computations

of the minimal AS needs are based on the stochastic

characteristics of ACE and its components, as deter-

mined in Step 1. Regarding the fact that some of the

AS are mutually substitutable (i.e., faster-responding

services such as secondary control can be substituted

for slower-responding services such as tertiary control),

Part G 65.2

1156 Part G Infrastructure and Service Automation

Table 65.3 Reliability indices based on area control error time series evaluation

Performance Limit Description

index

value(%)

rACE

3.2 Probability that the absolute value of the one-minute average of ACE exceeds L

=100 MW

rACE

4.6 Probability that the absolute value of the one-hour average of ACE exceeds L

=20 MW

rACE

0.027 Probability that the one-minute average of ACE is higher/lower than ±100MW for more than

15min

rACE

60t2

0.829 Probabilitythat the one-hour average of ACE is higher/lower than ±20MW within two consecutive

hours

rACE

60t3

0.140 Probability that the one-hour average of ACE is higher/lower than ±20MW within three consecu-

tive hours

rACE

60t4

0.040 Probability that the one-hour average of ACE is higher/lower than ±20MW within three consecu-

tive hours

the needs of the AS reserves are expressed in the form

of inequalities (65.14)–(65.18) rather than exact recom-

mended values for each AS category

RZSR

≥RZSR

min

, (65.14)

RZSR+RZTR

≥RZSR

min

+RZTR

min

, (65.15)

RZSR+RZQS+RZTR

≥RZ

(

min

, (65.16)

RZSR+RZTR

−

≥max



−

(

min

, RZSR

min

+RZTR

−

min



(65.17)

where RZ

(

min

≥ 0, RZ

−

(

min

≥ 0 are total mini-

mal volumes of positive and negative reserves, and

RZSR

min

≥0, RZTR

min

≥0 and RZTR

−

min

≥0 are the

minimal required volumes of the secondary, tertiary

positive and tertiary negative reserves, respectively. In

addition, (65.18) represents a UCTE N −1 criterion

RZSR+RZQS ≥C

N−1

, (65.18)

where C

N−1

is the installed power of the largest gener-

ator within the control area.

As the energy of the quick-start reserve is limited

and due to its quality in terms of a fast response after ac-

tivation, it should be preserved for future use asmuch as

possible. Thus, it is recommended that this fast service

should be replaced by a slower one whenever possible,

and the same amount of reserve should be planned for

such a slower service. This reasoning is expressed by

RZN

>30

=RZQS . (65.19)

Figure 65.9 illustrates how the total minimal vol-

umes RZ

(

min

≥ 0andRZ

−

(

min

≥ 0 are determined

from the cumulative distribution function of ACE

According to thereliability standards, theabsolute value

of the closed-loop ACE is allowed to exceed the thresh-

old 100MW in rACE

% of cases. If this is split sym-

metrically into positive and negative values, an ACE

that is higher than 100MW should be compensated

by the control reserves except in rACE

/2% of cases.

Hence, we need at least RZ

(

min

≥0andRZ

−

(

min

≥0

reserves to satisfy the standards of reliability derived

from area satisfactory performance in the past.

The secondary reserve RZSR

min

should compensate

for the fast variations of the open-loop ACE,andthe

tertiary reserve RZTR

min

, RZTR

−

min

should compensate

for the slow ACE variations.The minimal needs of these

services are determined in a similar manner to the to-

tal needs, but by utilizing the cumulative distribution

functions of ACE

OV slow

and ACE

OV fast

The values of the minimal AS requirements take

the form of a time series, which reﬂects the fact that

thebehavioroftheACE also varies; it is typically

more uncertain during transition periods with change-

able weather and temperature, such as during spring or

fall.

65.2.3 AS Bids from Generation Companies

Before calculating the costs of the optimal AS set, we

also need to know what energy and AS bids can be

expected on the market (Fig.65.7). The underlying as-

sumption is that under liberalized market conditions,

Part G 65.2

Securing Electrical Power System Operation 65.2 Ancillary Services Planning 1157

generation companies are only motivated by economic

factors, and their aim is to maximize their proﬁt re-

gardless of transmission system needs and reliability

issues. A generation company decides how much of

its installed capacity will preferably be offered in the

form of energy sold to domestic retailers or abroad and

what amount will be offered to the TSO in the form of

AS. This part of the model solves a mixed integer lin-

ear programming problem involving unit commitment.

The main model inputs and outputs are summarized in

Table 65.4.

The model does not search for a demand-bid equi-

librium on the market, because the computational time

associated with a year-ahead simulation would be too

high. Hence, the expected AS and energy prices are

entered as the inputs. In other words, the model tells

the TSO the amount of AS reserves that are ex-

pected to be available on the market if their prices

are given. Thus, for instance, the TSO can enter the

maximum acceptable prices of AS and see whether

or not the bids meet the minimal AS needs in such

a case.

65.2.4 Procurement of Ancillary Services

The key point is to optimize the AS selection cover-

ing the minimal AS needs based on the lowest costs for

AS reservation. In the case where the minimal needs

of AS in a certain category determined in Step 2 are

higher than the offer on the market, the optimal pur-

chase algorithm takes this limited offer into account

by substituting the AS with a different AS category

if possible. The optimal purchase is stated as a linear

optimization problem

min

∀k

RZSR

RZTR

−

RZQS

RZN

>30k



∀k



RZSR

RZTR

−

RZTR

−

RZQS

RZN

>30k

RZN

>30k



(65.20)

subject to the constraints (65.14)–(65.18). The index

k denotes individual AS providers, C

RZSR

, C

RZTR

−

, C

RZQS

, C

RZN

>30k

are bid prices of the re-

spective AS, and RZSR

, RZTR

−

, RZQS

RZN

>30k

are the AS volumes purchased from the kth

AS provider.

Table 65.4 Main inputs and outputs of the AS market

model

Model inputs

1. List of power producers acting on the market and portfo-

lio of their generation units

2. Technical data for generating units (type of technology,

minimal and maximal outputs, certiﬁed AS ranges, etc.)

3. Economic characteristics of generating units (variable

costs, start-up and shut-down costs)

4. Planned unit outages

5. Domestic power consumption and foreign exchange

forecast

6. TSO’s expert estimate of future energy and AS prices

based on knowledge of the current market situation and

possible future trends

Model outputs

1. Time series of expected volumes of AS offered by each

of the AS providers in the market

65.2.5 Availability

of Self-Regulation Power

The transmission system is balanced by not only the

TSO but also the area generators. The generation

companies are economically motivated to meet their

contracted production. Thus, if one or more of their

units fail, the remaining capacity of the other units

is activated, if possible, to compensate for the forced

outage (self-regulation). However, this ability varies in

time and depends on instantaneous domestic consump-

tion, foreign exchange, etc. Therefore, an algorithm

for computing time-variable available capacity for self-

regulation is implemented and conducted as Step 5 in

Fig.65.7. In principle, the remaining available capacity

can be determined as

= P

−P

OUT

−P

, (65.21)

where P

is the installed capacity of generation com-

panies, P

OUT

are planned outages, P

is the sold energy

and P

is sold AS. The self-regulation ability is then

taken into account in Monte Carlo simulations where

the forced generation unit outages are randomly gener-

ated.

65.2.6 Monte Carlo Simulation

of Power Balancing

The quality of the transmission system operation must

be validated for the intended ancillary services acquired

inStep4inFig.65.7 for several reasons. First, only two

of the performance indices rACE

and rACE

eval-

Part G 65.2

1158 Part G Infrastructure and Service Automation

uating the ACE time series were taken into account

when calculating technical needs of the AS. Second,

purchasing the AS may result in a different composition

of AS due to the limited offer from the AS providers.

To learn more about the behavior of the transmission

system, a power balancing control system simulator is

used to simulate the automatic activation of AS and to

mimic human operator interventions at the TSO control

center [65.19].

The logic used by the TSO operator manually ac-

tivating the respective categories of AS is modeled

as if-then rules. The operator evaluates past and pre-

dicted values of ACE, considers the balancing power

reserves left for immediate and near-future use and the

dynamics of the respective AS (i.e., the delay, ramp

up, ramp down), and checks the time taken for the ac-

tivation or deactivation of the AS to be realized. By

means of Monte Carlo simulations,this model generates

time series for open-loop ACE [65.19], AS activa-

tions/deactivations (both automatic and operator-based)

and, consequently, closed-loop ACE during a planning

period when using the recommended set of AS. Addi-

tional nonguaranteed types of AS,suchasemergency

assistance from abroad and the purchasing of balanc-

ing energy are also used in simulation runs. These types

of services are characterized by zero reservation costs

and limited availability (i.e., the services might be tem-

Reserved capacities: RZSR, RZTR

, RZTR

–

, RZQS, RZN30

RZSR

Man

Auto

Quick-start reserve

Tertiary reserve

Standby reserve

Emergency assistance from abroad

Dispatch rules

Energy management

Contingency management

Setpoints

for utilities

Response

dynamics

ACE

Open-loop ACE

without outages

ACE

= Remaining ACE not compensated by the TSO = P

: Forced outages compensated by utilities; ΔP

= Emergency assistance (from abroad)

Self-regulation capacity

Forced outages

Balancing power

applied

Secondary controller

AGC

ACE

predictor

ΔP

ACE

Area generation

and load pool

ACE

Self-regulation

utilities

Fig. 65.10 Block diagram of a power balancing control system simulation for the area

porarily unavailable when required by the operator for

balance control). The availability of the nonguaranteed

types of AS is a parameter that can be set in the simula-

tor.

The master block diagram shown in Fig.65.10 gives

an overall picture of the simulator used in Step 6 in

Fig.65.7. The simulator is directly related to the feed-

back control scheme from Fig.65.3.

Simulated ACE

, ACE with automatic activation

of the secondary reserve and manual activation of the

tertiary reserve, quick-start and standby reserves, emer-

gency assistance from abroad, and balancing energy

from abroad are shown in Fig.65.11. The simula-

tion shows a record for two days with a large power

deviation originating from an outage of a 1000MW

generator. The area control error not compensated by

the TSO ACE

is the curve with the highest mag-

nitude. The producer compensates for the unintended

power deviation by activating its own power reserves.

As a result, ACE

tends to be drawn towards zero but

the self-regulating action is not efﬁcient enough and so

a signiﬁcant power deviation remains for about 12 h,

which is unacceptable.

The situation improves when the TSO activates its

reserves. The secondary controller responds to the sud-

den increase in ACE almost immediately (red). As

the secondary power reserve available at that time

Part G 65.2

Securing Electrical Power System Operation 65.2 Ancillary Services Planning 1159

ACE

RZN

>30

Emergency

from abroad

25 30 35 40 45 50 55 60 65 70

Day n Day n +1

(MW)

(h)

1000

800

600

400

200

–200

–400

–600

Fig. 65.11 Activation of ancillary services for area power balancing

is about 400MW, it is not possible to cover the

massive outage solely using this reserve, and other

services must be called up to assist. The TSO’s op-

erator activates the quick-start reserve (cyan) and

the tertiary reserve (magenta) at the same time. The

pumped-storage quick-start must be saved as its ca-

pacity is limited, so the operator tends to replace it

as soon as possible with the services with a longer

response time, the standby reserve (yellow), and/or

tries to acquire services that are not guaranteed: emer-

gency assistance from the adjacent area (green). As

a result, the ACE (black) is minimized and the mas-

sive outage is well compensated for via a temporary

combination of ancillary services, or rather their ac-

tivation. The TSO’s manual intervention also assures

that the capacity of the secondary reserve is restored

as quickly as possible so the output of the auto-

matic load-frequency controller is driven away from

the limits, the power reserve purchased by the TSO in

advance.

65.2.7 Control Performance Evaluation

Although the AS needs were determined in Step 2 with

the aim of meeting the required reliability in terms of

the indices rACE

and rACE

, not all of the phe-

nomena inﬂuencing the reliability could be taken into

account inthe analyticalcomputations. For instance,the

dynamics of AS activation/deactivation and the limited

capacity of pumped-storage reservoirs are neglected.

Moreover, the AS set recommended for purchase is

often different from the minimal set due to limited of-

fers or the substitution of one AS for another. Hence,

through the statistical evaluation of the Monte Carlo-

simulated ACE, more realistic values of the reliability

indices rACE

and rACE

can be determined, and the

values of the other ﬁve indices introduced in Table 65.3

can also be found.

Other performance indices can be evaluated, includ-

ing NERC’s CPS2 and DSC, with the exception of

CPS1, as it requires the tie-line frequency deviation to

be simulated whereas the simulator [65.17] does not

support frequency.

65.2.8 Case Studies

Algorithms of AS planning and the area control system

simulator can be used for various studies and analyses

supporting TSO’s decision-making when planning and

purchasing ancillary services.

Technical and economic aspects are closely related

in this task, and it is essential to answer the question of

Part G 65.2

1160 Part G Infrastructure and Service Automation

how much it will cost to guarantee a commonly agreed

and accepted level of reliability in the system services

provided by the TSO. The methodology presented here

enables the relation between the reliability measured

through statistical evaluation of ACE and the cost of

purchasing the power reserves to be studied. A typical

relation between two reliability indices and the cost of

AS is shown in Fig.65.12.

The plot shows the relative increase or decrease in

the AS purchase cost when the request for reliability,

measured as the relative number of cases when minute

averages and hourly averages of ACE exceed speciﬁed

levels (e.g., 100MW for a minute and 20 MW for an

hour), is relaxed.

Several studies illustrating the capabilities of the

method can be found in [65.16], where the results of

year-ahead planning based on specifying the needs for

the individual services and the recommended optimal

purchase are shown. There can be signiﬁcant differ-

ences between the needs and the purchase, which is

due to the fact that RZQS availability is limited to ex-

isting pumped-storage plants, and thus inﬂuenced by

their scheduled outages. A similar rule applies to the

standby reserve RZN

>30

. Thus, more positive reserves

must be acquired in the category of faster-response

AS purchase

cost change (%)

Acceptable

Not acceptable

Low reliability

Acceptable

Standard: rACE

= 3,2%

rACE

= 4,6%

rACE

(%)

rACE

(%)

8.5

–8.5

–17

4.6

3.2

Nominal cost

Fig. 65.12 AS costs as a function of required reliability

services, according to (65.16). The expert-entered

prices of RZSR, RZTR

and RZTR

−

were related as

RZSR

> C

RZTR

−

, so the symmetrical RZSR

service is preferred to tertiary control, which results

in low or even zero purchasing of tertiary reserves

RZTR

−

. However, the volume of RZTR

is higher than

the needs during some periods of the year, as it substi-

tutes for the lack of RZQS andRZN

>30

in order to reach

the minimal positive reserve RZ

(

min

. A correlation be-

tween greater amounts of faster-response services and

improved reliability of power system operations can

also be observed.

Self-regulation of power producers is a risky fac-

tor. The level to which a generation company is willing

to eliminate the difference between the planned and

the actual production also depends on the penalty the

company will have to pay for the internal error. Stud-

ies carried out on the power balancing control system

simulator can be used to ﬁnd out the acceptable rela-

tion for internal error charges. Reducing internal power

balancing reserves of the power producer will end up

increasing the cost of reserves that need to be acquired

for the entire area by the TSO. This cost will then be re-

lated to the reductionin the producer’s balancingpower.

Simulation studies in [65.16] show that a lack of self-

Part G 65.2

Securing Electrical Power System Operation 65.2 Ancillary Services Planning 1161

regulation capability in the utilities could increase the

TSO’s AS expenses signiﬁcantly.

The methodology presented deals with the TSO’s

ancillary services planning and purchasing in order

to prevent undesirable situations in the transmission

system under changing market conditions and with pre-

dicting the inﬂuence of these changes on the TSO’s

expenditure. The AS planning method is shown for

year-ahead planning (including various analyses and

sensitivity tests, e.g., costs–reliability relation), and the

results provide support for the TSO’s strategies in the

annual AS auction. Despite the fact that the proposed

concept is tailored to the needs of a particular control

area in the UCTE interconnection, it represents a com-

pact approach to optimal AS planning.

The presented planning and modeling approach is

based on historical data. However, the method can in-

corporate the characteristics of new types of generation

units, study what installing wind turbines would mean

for ancillary services planning, and evaluate the cost of

reaching the same level of reliability.

Price elasticity can be incorporated into the model

of energy and AS bids in order to make the entire model

more realistic. The costs for AS will then supposedly

increase if the (optimal) volumes required in a certain

AS category are higher than usual.

The challenging task is todetermine the value of un-

served energy in the area. The value of unserved energy

is the loss to the economy if a MWh of the energy re-

quired by consumers cannot be supplied. Consequently,

the described methodology allows the optimal balance

between the costs of AS and the costs of unserved en-

ergy to be found.

65.2.9 Related Topics

Generation from Renewable Resources

Energy sources are undergoing permanent restructur-

ing. Large-scale renewable resource installations like

wind power, which is the world’s fastest-growing en-

ergy technology, represent a new challenge to the power

system. The ﬂuctuating nature of wind power affects

power system reliability, deviating from the planned

powergeneration, whichleads to power balancing prob-

lems. Selecting appropriate approaches for calculating

the wind power-induced balancing power reserve while

respecting the speciﬁcity of regulatory and market sit-

uations in the region is currently an issue [65.20, 21].

Several detailed technical investigations of the impacts

of wind power plants on grid ancillary services needs

have recently been performed [65.22]. Although the

approaches vary, three utility timeframes appear to be

the most signiﬁcant issues: regulation, load following

and unit commitment. This article describes and com-

pares the analytic frameworks from recent analysis and

discusses the implications and cost estimates of wind

integration. The ﬁndings of these studies indicate that

relatively large-scale wind generation will have an im-

pact on power system operation and costs, but that these

impacts and costs are relatively low at the penetration

rates that are expected over the next few years.

Transmission Congestion Management

Grid management has become notably more complex

over the past few decades. The level of grid usage is

increasing markedly in every region. Consistently high

utilization of critical assets in a network often means

that these assets become bottlenecks that limit the use

of the network and its ability to serve demands at ev-

ery level consistently and reliably. These points on the

grid create transmission constraints. In some cases they

create congestion by limiting the buyers’ ability to se-

cure energy from the most economical source, meaning

that energy must be purchased from closer, more costly

generators. In other cases, transmission constraints can

cause a reliability problem when customer demands ex-

ceed the delivery capabilities of the transmissionsystem

plus local generation.

As the power ﬂows over a grid with rapidly chang-

ing electrical topologies increase, the local utility is no

longer able to see and manage every factor affecting

the ﬂows through its share of the transmission grid.

The scope of grid monitoring and control that worked

effectively in thedays of highcapacity margins and lim-

ited interutility ﬂows are no longer sufﬁcient. Higher

grid usage causes greater reliability challenges, and

policy-makersand utilities then come to the opinion that

increasing the sizes of grid oversight and control orga-

nizations is the only effective way of dealing with these

challenges.

This problem is of great importance in coun-

tries/areas where, besides energy and ancillary services,

transmission capacity is a commodity in a deregulated

market for electricity and/or where the transmission ca-

pacities are almost/temporarily saturated (as is the case

when the transmission systems were not designed for

the power ﬂows required).

The problems of ancillary services planning and

real-time balancing power reserve activation discussed

in this chapter do not address one important issue: the

limited capacity of the transmission lines. It can hap-

pen that the called-up ancillary service will overload

Part G 65.2

1162 Part G Infrastructure and Service Automation

a particular transmission line when there is a sched-

uled energy exchange. As the approach described so

far models the control area as a pool, it does not al-

low the control area network topology to be taken into

consideration. Other approaches must be applied.

Reference [65.23] presents major congestion man-

agement methods in the new market environment from

the international technical literature, which are pre-

sented and classiﬁed according to security-constrained

generation redispatch, zonal/cluster based management,

a network sensitivity factor-based method, congestion

management using FACTS devices, congestion pricing

and market-based methods, congestionmanagement us-

ing demand-side resources, and ﬁnancial transmission

rights.

In a restructured power system, the transmission

networkis thekey mechanism forgenerators to compete

in supplying large users and distribution companies.

The role of the competitive electricity markets is de-

scribed in [65.24]. The lessons learned by ERCOT

(Texas ISO) in relation to ﬂow-based zonal redis-

patch to relieve transmission congestion are discussed

in [65.25].