A Concept for Reliability Assessment for the Provision of Ancillary

Services

Marita Blank

and Sebastian Lehnhoff

Department of Computing Science, Carl von Ossietzky Universit¨at, Oldenburg, Germany

OFFIS, Institute for Information Technology, Oldenburg, Germany

Keywords:

Reliability, Ancillary Services, Smart Grids.

Abstract:

In order to ensure the quality of supply in the electric energy system the ancillary services frequency and

voltage control are necessary. With regard to the ongoing decentralization of power supply, approaches are

being investigated to provide those ancillary services by decentralized units some of which are highly volatile

in power generation. This leads to the question of the reliability with which ancillary services can be provided.

In this paper a method is proposed that allows for the assessment of reliability of aggregations of distributed

units with respect to their ability to provide ancillary services.

1 INTRODUCTION

The penetration of renewable electricity generating

units in the electrical supply system steadily increases

and with that large power plants are replaced. In ad-

dition, more ﬂexible and controllable loads are con-

nected to the grid. This leads to new challenges and

increased demand for ancillary services. Ancillary

services are services to support a stable and secure

operation of the network and need to be constantly

available. These services include frequency and volt-

age control.

To a great extend ancillary services have to be pro-

vided by distributed ﬂexible consumers and produc-

ers since large power plants are no longer available

to the extend as they are today. A variety of devices,

e.g. small power plants, batteries, and house hold ap-

pliances being captured under the term unit can be

controlled by intelligent agents. However, individual

units usually can only provide a small contribution to

ancillary services. Furthermore, from today’s point of

view units must be able to reliably provide ancillary

services in the sense that they are 100 per cent avail-

able. The paradigm proposed in (Nieße et al., 2012)

allows the pooling of units in order to provide ancil-

lary services in a decentralized and distributed way.

The resulting coalitions must be able to reliably pro-

vide ancillary services such that safe operations are

guaranteed. For this reason, it will be necessary to

evaluate the reliability of ancillary service provided

by coalitions of distinct units.

In this paper a method is proposed for the proba-

bilistic evaluation of reliability of unit coalitions with

regard to the provision of ancillary services. There-

fore, a new understanding of reliability is introduced.

To this end, the ancillary services frequency and volt-

age control are discussed in section 2. Furthermore,

a generic description of those services is given. In

the following section 3, the term of reliability is dis-

cussed and a proper deﬁnition in this context is intro-

duced. A concept for a method for the evaluation of

a coalition’s reliability is presented. The paper ends

with conclusion and outlook in section 4.

2 ANCILLARY SERVICES

In order to maintain service quality in the electri-

cal supply system, amongst others the aspects of fre-

quency and voltage quality must be kept. This means

that operation constraints must not be violated. To

this end ancillary services are applied.

Section 2.1 discusses the system boundaries of

frequency and voltage band. Furthermore, a deﬁni-

tion of the ancillary services frequency and voltage

control as well as their properties is given. Section

2.2 follows with a formal description of these two an-

cillary service types. In section 2.3 an outline is given

of how ancillary services are provided by agent coali-

tions.

123

Blank M. and Lehnhoff S..

A Concept for Reliability Assessment for the Provision of Ancillary Services.

DOI: 10.5220/0004408601230126

In Proceedings of the 2nd International Conference on Smart Grids and Green IT Systems (SMARTGREENS-2013), pages 123-126

ISBN: 978-989-8565-55-6

 2013 SCITEPRESS (Science and Technology Publications, Lda.)

2.1 Deﬁnition and Properties

In IEC 60050-617 (system) ancillary services are de-

ﬁned as “services necessary for the operation of an

electric power system provided by the system opera-

tor and/or by power system users” (see (IEC, 2011)).

In this paper, the ancillary services frequency and

voltage control are considered. In the sequel, a brief

overview of network frequency and voltage band are

given, respectively, since they constitute margins for

safe grid operations. Moreover, both ancillary ser-

vices are described, reﬂecting the actual situation.

The network frequency is a global variable that

has the same value in the whole system and should be

close to the setpoint value of 50Hz in the European

power grid. However, a deviation of the frequency

from this target value cannot be avoided. Neverthe-

less, the frequency must stay within a certain range.

In order to keep the frequency at a stable level, it

is necessary that the generated and consumed active

power is in balance. To ensure this, frequency control

reserves must be provided. This reserve is differenti-

ated into primary, secondary and tertiary control that

are triggered depending on a certain deviation from

the set point and have to be delivered within different

time intervals. For details see e.g. (Rebours et al.,

2007)).

Another crucial variable is the voltage at each grid

node. The target value to keep the voltage at is the

nominal voltage of a voltage level. Due to changes in

the power feed-in and consumption along a line the

nodal voltages usually deviate from the nominal volt-

age. However, not more than a certain ration from the

nominal value is allowed, e.g. in the distribution grid

a deviation of not more than ten per cent is allowed.

Since voltages differ from node to node it is nec-

essary to provide voltage control locally. This is done

by injection or consumption of reactive power accord-

ing to given (by e.g. the TSO) characteristic curves.

Additionally, in the distribution grid a combination of

active and reactive power for regulating the voltage

can be applied.

2.2 Formal Description

The ancillary services described before have common

attributes. This allows for a generic description of

ancillary services of different types. The following

speciﬁcation covers the services types of frequency

and voltage control, but it can be extended and added

to describe other service types as well.

Agent coalitions provide a certain amount of an-

cillary services (see 2.3). In order to determine

whether ancillary services and in which amount are

necessary, the considered quality aspect (frequency or

voltage level) must be investigated. To this end, there

are three main properties of a quality aspect. First of

all, a measure of the quality aspect must be available

in order to evaluate if ancillary services must be acti-

vated, termed the measurable quantity. This variable

has a set point. Furthermore, the value of the measur-

able quantity is allowed to vary within a feasible area.

The deadband is the area within which the quantity is

allowed to vary in without any reaction of the system.

As mentioned before, the measured value can

be inﬂuenced by providing a certain quantity, here

termed ancillary service quantity. In case of fre-

quency this is active power, in case of voltage this

is reactive power, or (as voltage control is becoming

more important in the distribution grid) a combination

of active and reactive power, i.e. it can be a real or a

complex number.

In order to guarantee the provision of ancillary

services with the necessary amount, in the presented

concept ancillary service products are introduced that

allow for the annotation of costs for the provision of

ancillary services. An ancillary service product is

deﬁned as a triple of a time interval – the product

horizon, ancillary service amount, and corresponding

costs.

2.3 Provision of Ancillary Services

The system has certain demands on ancillary ser-

vices regarding its amount and a minimum reliability

(actual requirements depend on the ancillary service

type). Hence, the system calls for bids for the pro-

vision of those services for a given time interval, the

product horizon. The units or agents (see below for

details) that want to contribute to a service provision

determine what amount they can provide. If this con-

tribution is not sufﬁcient, agents must negotiate with

other agents in order to form coalitions that fulﬁl sys-

tem requirements. In case a coalition is chosen by the

system to provide an ancillary service this coalition

is responsible that the required amount is available

with a required reliability throughout the whole prod-

uct horizon. If an ancillary service is actually called

by the system the responsible coalition must deliver

the required amount automatically.

A unit is an electrical device or a set of electrical

devices that produces or consumes electric power and

is further equipped with an embedded system. A unit

is associated with exactly one grid node. An agent is a

software component exhibiting intelligent behaviour.

Each agent represents a unit, hence there is a one to

one relationship between units and agents. A coali-

tion is an aggregation of units resp. agents.

SMARTGREENS2013-2ndInternationalConferenceonSmartGridsandGreenITSystems

124

Each agent must calculate the amount of ancillary

service quantity its unit can contribute to an ancillary-

service coalition. The contribution of agent and corre-

spondingly its represented unit within a product hori-

zon is deﬁned as the set of pairs consisting of the

amount of ancillary service quantity and costs for all

product intervals (assuming that a product horizon is a

union of product intervals). For conceptual clarity, in

the following, the focus is laid on the ancillary service

amount contributed rather than the costs it causes.

Many of the decentralized units are subject to a

certain behaviour of users (e.g. controllable loads or

combined heat and power plants), or depend on un-

certain and volatile weather phenomena, such as pho-

tovoltaic or wind turbines. Consequently, planning

of available amounts of an ancillary service quan-

tity strongly depends on predictions which inherently

bear errors. Thus, coalitions that provide ancillary

services must be evaluated with regard to reliability.

3 CONCEPT OF RELIABILITY

As pointed out before, due to a decentralized provi-

sion of ancillary services, i.e. by agent coalitions, an

assessment of reliability with which these coalitions

are able to provide ancillary services is necessary. In

section 3.1 a deﬁnition for reliability is discussed and

a proper deﬁnition is given. In section 3.2 a concept

for reliability evaluation is introduced.

3.1 Deﬁnition and Evaluation

of Reliability

In (IEC, 2011) reliability of an electric power system

is deﬁned as “the probability that an electric power

system can perform a required function under given

conditions for a given time interval.” It is stated that

“reliability quantiﬁes the ability of an electric power

system to supply adequate electric service on a nearly

continuous basis with few interruptions over an ex-

tended period of time”. To this end it is necessary

that ancillary services are available at a rate of 100 per

cent. If ancillary services are being provided by coali-

tions consisting of distributed units this requirement

can hardly be met since those units may be volatile in

production or consumption.

A prerequisite of reliable system performance is

the reliable functioning of its components. The re-

liability performance of an item according to (IEC,

1990) is deﬁned as the ability to “perform a required

function under given conditions for a given time in-

terval”. It is further stated that in some applications

an appropriate measure for reliability performance is

expressed by a probability. In order to guarantee a re-

liable supply of ancillary service by distributed units,

a metric is necessary that allows the assessment of the

performance of unit coalitions regarding their ability

to provide ancillary services beforehand. To this end,

the following deﬁnition is proposed:

Reliability of a coalition with respect to the

provision of an ancillary service product is the

probability with which this product is avail-

able within a product horizon under certain

conditions.

This deﬁnition of reliability can be used in the course

of coalition formation as it can be integrated in the

objectivefunction of ancillary service coalitions given

a minimum acceptable reliability value depending on

the ancillary service type of interest. In the following,

a corresponding concept is introduced which allows

the calculation of reliability.

3.2 Hierarchy of Reliability Evaluation

In this section a concept for a method is introduced

for the evaluation of agent coalitions with regard to

their ability to provide an ancillary service product.

The reliability of a coalition with regard to the pro-

vision of an ancillary service product depends on the

reliability of all of the coalition’s member units. The

reliability of a unit again is inﬂuenced by several fac-

tors. This leads to a hierarchy for the evaluation of

a coalition’s reliability. In the following, the non in-

ﬂuenceable and inﬂuenceable factors are introduced

in detail. Based on these factors, the reliability of a

unit can be calculated. Finally, the determination of a

coalition’s reliability is discussed.

Non inﬂuencable factors that have to be taken into

account for reliability evaluation are for instance pre-

dictions of the operation behaviour that might depend

e.g. on weather phenomena or user proﬁles. But also

factors such as unplanned outages must be regarded.

In order to represent a unit, its agent must know

the unit’s behaviour given as an operation schedule.

The operation schedule and with that the resulting op-

eration ﬂexibility depend on predictions of meteoro-

logical phenomena, user behaviour and so on. Pre-

dictions deal with uncertainties and thus usually are

imprecise and prone to errors. Furthermore, errors

may occur with growing considered time horizons.

This is especially the case for generation units that

depend on the availability of certain resources, e.g.

wind or solar power plants. The probability that the

predicted power amount is actually fed-in is assumed

to decrease whereas the probability for both positive

and negative deviations increases with time, i.e. the

prediction error increases.

AConceptforReliabilityAssessmentfortheProvisionofAncillaryServices

125

There are methods to integrate several inﬂuences

in order to obtain a common model for prediction er-

rors such as convolution. This is amongst e.g. a com-

mon procedure for the determination of power reserve

dimension, as e.g. in (Kays et al., 2010). Given a dis-

tribution function from a (possibly convoluted) den-

sity function of the prediction error it is possible to

derive an available amount of ancillary service quan-

tity with a certain reliability. The other way around,

given a reliability level the appropriate amount of an-

cillary service quantity can be determined.

Given the prediction of a unit’s behaviour based

on non inﬂuenceable factors, inﬂuenceable factors

can be adjusted in order to achieve a certain goal, for

instance satisfy a given reliability level. In this paper

three factors of that kind are considered. Those are

the accepted level of reliability, the provided amount

of ancillary service quantity, and the considered pre-

diction horizon.

As mentioned before, the contribution of a unit to

an ancillary service product might in general be too

small to satisfy system requirements. Moreover, the

reliability of the contribution might not reach a re-

quired level. As a consequence, units form coalitions

with the goal to provide a required amount of an an-

cillary service quantity with a minimum value of re-

liability. The reliability value for a coalition’s contri-

bution must be determined based on the reliability of

each of its member units.

Here the three inﬂuenceable factors mentioned

previously also apply, i.e. units can adjust their con-

tribution, their individual reliability or the prediction

horizon can be adjusted. Moreover a forth factor con-

tributes to a coalition’s reliability evaluation - the size

and structure of a coalition. Often different units are

subject to the same inﬂuences, e.g. photovoltaics in-

stalled on neighboured houses. Hence, simultaneous

behaviour reﬂected by error distributions cannot be

modelled using convolution since stochastic indepen-

dence is required for its application. Suitable methods

for estimating coalition reliability will be developed

in future work. The estimation of a coalition’s relia-

bility can then be integrated into a coalition formation

strategy as part of an objective function.

4 CONCLUSIONS

AND OUTLOOK

In this paper a deﬁnition of reliability was introduced

for the assessment of agent coalitions with regard to

provision of ancillary services. Furthermore, a con-

cept was proposed that allows an ex-ante evaluation

of reliability. This method is applicable for differ-

ent types of ancillary services since those types have

common properties. In order to determine the relia-

bility of a coalition, ﬁrst the reliability of its member

units must be calculated. This is based on different

factors which were categorized into inﬂuenceable and

non inﬂuenceable factors. Based on the reliability of

its member units, the reliability of a coalition can be

determined. A coalition’s reliability depends on the

size and structure of the coalition. This procedure can

be applied in the course of coalition formation.

In future work the proposed techniques will be

extended and further techniques will be evaluated.

Especially methods from other domains such as ﬁ-

nance and insurance will be investigated and if pos-

sible adjusted to the considered context. The devel-

oped method will then be checked for validity, e.g.

using Monte Carlo Simulations. Further intended re-

search will also include costs and extend the existing

methods for the assessment of coalition reliability. In

particular, it must be taken into account if the same

inﬂuences on units and corresponding agents lead to

stochastically dependent behaviour within a coalition.

ACKNOWLEDGEMENTS

The Lower Saxony research network ’Smart Nord’

acknowledges the support of the Lower Saxony Min-

istry of Science and Culture through the Niederschsis-

ches Vorab grant programme (grant ZN 2764).

REFERENCES

(1990). IEC 60050: International Electrotechnical Vocab-

ulary - Chapter 191: Dependability and quality of

service. Technical report, International Electronical

Commission.

(2011). IEC 60050: International Electrotechnical Vocab-

ulary - Part 617: Organization / Market of electricity.

Technical report, International Electronical Commis-

sion.

Kays, J., Schwippe, J., Waniek, D., and Rehtanz, C. (2010).

Multidimensionales verfahren zur bestimmung des

regelleistungsbedarfes unter ber¨ucksichtigung von

unsicherheiten. Zeitschrift fr Energiewirtschaft,

34(4):267–278.

Nieße, A., Lehnhoff, S., Sonnenschein, M., Uslar, M., and

Appelrath, H.-J. (2012). Agent-Based, Self-Organized

Provision of Active Power and Ancillary Services in a

Smart Grid: a Use Case. pages 1–5.

Rebours, Y., Kirschen, D., Trotignon, M., and Rossignol, S.

(2007). A Survey of Frequency and Voltage Control

Ancillary Services - Part I : Technical Features. IEEE

Transactions on Power Systems, 22(1):350–357.

SMARTGREENS2013-2ndInternationalConferenceonSmartGridsandGreenITSystems

126