You are on page 1of 17

·-

'
DETERMINATION

OF

GENERATOR INERTIA

' BY

J. L. Gordon, P. Eng.

MONTREAL ENGINEERING COMPANY LTD.

Presented to.
Eastern Zone Meeting
Hydraulic Power Section

, Canadian Electrical Association


Halifax, 24th January 1961.
• -1-

l. INrRODUCTION

This paper deals with some of the considerations necessary in the selec -

tion of the rotating inertia required to be built into a hydro-electric

generating unit, so that the unit will perform satisfactorily under the

operating conditions for which it was designed.

Several analysis of this problem have been made in the past, the most

notable being the NEMA publication No. HT-4-1958, from which the gen-

erator inertia may be selected depending on the _unit size and the length

of the penstock. However, to our knowledge, no paper has been published

• which shows in a simple form the experience gained from past installations •

This paper will present on a chart similar to that produced by NEMA, a

tabulation of our experience in the selection of generator inertia for

varying conditions, as determined from the characteristics of about 40

Kaplan, propellor and Francis turbine installations.

2. COST OF INERTIA

As the benefits of power are brought to outlying communities by means of

small isolated hydro developments, a study of the generator inertia

requirements is necessary since it involves a compromise of conflicting

requirements. Generally speaking, the more inertia that is built into a

generator the better, since it will contribute towards a stable system

with good frequency regulation. However, in small isolated developments,


costs have to be kept to a bare minimum, and since extra inertia invariably

adds cost to the project, a compromise must be reached wherein the iner-

tia is selected consistent with economy and good frequency control.


;

• -2-

An illustration of the extra costs involved from adding inertia to the gen-

erator may be obtained from a common rule of thumb which states that the

cost of a generator will be increased by 1% for every 4% increase in

inertia. If a large inertia increase is required, both the size and weight

of the rotor must be in~re~sed. A larger rotor generally means an in-

crease in powerhouse area to accommodate the larger unit, and the in-

creased rotor weight will require a crane of larger capacity. Thus,

while the cost of adding inertia may, at first glance, appear to be low,

the actual cost, once all the factors have been taken into account, may

be considerably higher. Hence, it is essential that the generator

• inertia be kept to a minimum.

3. USE OF GENERATOR "H" VALUE

The common measure of generator inertia is the H factor, which is:

. 2. ~) I
l-t

Where WR2 is the generator inertia in foot-pound units

N is the rpm

KVA is the generator rating

H is the inertia constant, and is expressed in Kilowatt -

seconds per KVA. It is used to give an approximate value of the fre-

quency change due to a sudden step load change assuming no reaction

• from the governor,and using the rule that a 1% frequency change will

occur with a load change equal to 2% of the generator H value.

ample, assume a generator rated at 40,000 KVA with an H value of 2.5,


For ex-
• -3-

then the total Kilowatt - seconds in the flywheel effect of the generator

will be 100,000. If there is a sudden load increase of 5000 KW applied

for 2 seconds, the load change would then be 10,000 KW seconds, or 10%

of the generator flywheel inertia. This will then give a frequency drop

of one-half of the percentage load change or 5%, that is, 3 cycles in a

60 cycle system.

The use of the generator H value is particularly useful when applied to

a system with several generators. In this case, the total KW seconds of

flywheel effect are simply added together, since the speed of all units

• will rise or fall together with the system frequency. For example,

assume a generator of 10,000 ¥NA with an H value of 2, and one of 20,000

KVA with an H value of 2.5, the total flywheel effect would be 70,000 KW

seconds. A load drop of 3500 KW for 2 seconds would then produce a speed

rise of 5%.

The above simple formula neglects to take into account two of the major

factors affecting speed change, namely, governor time and inertia in the

system, both of which are discussed later.

4. INCREASING GENERATOR INERTIA

For a given generator KVA rating, there would appear to be two means of

increasing the H value, by increasing the speed or by increasing the

inertia WR2 • However, the speed of a unit is fixed by the hydraulic

• design conditions and is usually the highest which can be tolerated,

since, generally speaking, the higher the speed the lower ls the turbine

cost. Also, a higher unit speed usually reduces the normal inertia
• -4-

required for the KVA output, so that the product of inertia by speed

squared, is constant for a given generator rating over a wide range of

speeds (about 100 to 400 spin). Fig. 1 shows the value of the standard

generator i nertia constant H plotted against the generator rating. As

would be expected, there is a gradual increase in the generator H value

as the size of the unit increases.

Hence, for any one size of unit, the only method of increasing the inertia

is either to increase the weight W or the radius of gyration R or both.

As mentioned previously, any increase in these factors may affect the

• s.
design of t he powerhouse and its crane •

FACTORS AFFECTING GENERATOR INERTIA

Advances in the design of generators over the years have produced units

which have a lower normal inertia than a comparable generator manu-

factured, say 20 years ago. This has produced a unit which is admira -

bly suited for _insta llation in a large system, but somewhat lacking in

inertia when used in a small isolated system.

Thus the study of inertia on a small system is important, since a

considerable saving in cost can be obtained if the bare minimum of iner-

tia is insta lled , instead of an arbitrarily selected inertia which

could be overly conservative.

• The factors which must be taken into account when making this study are:

(a)

(b)
Size of the system to which the generator will be connected

The allowable f requency fluctuation


• -5-

(c) Length and shape of water passages

(d) Governor time

(e) Type of load

To discuss each of these in turn:

(a) Size of System

In a large system which has good frequency control, the stan-

dard generator inertia will usually be sufficient, since any

increase above normal in the ine r ti a will have only a neg-

ligib l e e ffect on t he system frequency band width,

• However, suf ficient ine r tia must be installed to keep the speed

rise on full load rejection to within a reasonable figure. As

mentioned prev iously , larg e speed rises affect the design of

the generator.

For small isolateg systems and even large systems with poor

frequ ency control, careful selection of inertia must be made,

with the inertia s tudy becoming more and more important as the

capacity of t he proposed generator increases with respect to

th e total system capacity.

The selection o f criteria to determine whether a system is

large or smal l is somewhat complicated. In general, any unit

• on a system which would be supp lying 40% or more of the total

system i nertia , shoul d be designed as an isolated unit. The


• -6-

relationship of the system inertia to the load change should

also be compared. Any load change which in KW seconds, is 10%

or more of the system KW seconds of inertia should be analyzed

to determine its effect on frequency.

Also, allowance should be made in the design of the generator

for abnormal conditions on the system. If, for example, the

proposed insta llation could become isolated from the system

through a transmission line fault, and remain on line supplying

a local load, the generator inertia would have to be checked to

• see if it was sufficient to provide a reasonably satisfactory

service during this fault period. During this fault period,

one would assume that a frequency band width of one-half to

one cyc l e could be tolerated, with perhaps about one-tenth of

one cycle being the frequency band width during normal operation.

(b) Allowable Freguency Fluctuation

The permissible frequency band width will depend on the type

of load. For machines requiring very close conl!Tol of fre-

quency such as in paper making, a frequency change of only

one-tenth of one percent may break the paper roll. On the

other hand, mining equipment can usually function satisfactorily

with a frequency band of 2 or even 4 cycles. Normally, it


should be possib le to maintain frequency within a band width

of about 2 cycles on a small system with a mining load. Care


• -7-

should be exercised in the selection of band width criteria,

since the smaller the desired band width, the greater the

required inertia and hence, the greater will be the cost of the

development.

(c) Water Passages

The effects of water hammer in the supply pipe to the turbine

must also be considered. Generally, the longer the pipeline

the greater the water hammer and the greater will be the re-

quired inertia t o maintain the desired frequency band. This


is due to the fact that on a load rejection, water hammer will

momentarily Increase the effective head on the · unit, and

since the horsepower is proportional to the three-halves

value of the head, the effects are compounded. Thus, a water

hammer which increases t he nett head by 30% will momentarily

increase the turbine horsepower output by about 48%. This

large excess of energy, produced by the turbine, is absorbed

by the flywheel effect of the generator by a rapid Increase in

speed. The opposite happens on a load acceptance causing a

negative water hammer. Thus, the effects of water hammer con•

tribute towards instabili ty in governing.

Another factor which could limit the rate of gate closure is the


effect of negative water hammer in the draft tube. The wicket

gates should be closed at a rate which will not cause rupture

of the draft tube water column. This, however, will only apply
• -8-

in large low head units with short penstocks.

There is, however, an alternative to increasing the generator

inertia to obtain good frequency control. By increasing the

diameter of the pipeline, and thus reducing the water velo-

city, the governor time may be shortened for the same water

hammer effect. This reduces the overspeed on a full load

rejection.

Also, lower water velocities permit use of a faster dashpot

return time. The net effect on a small load change_ is to

• slightly reduce the frequency variation. However, the effect

of lower water velocities on frequency are usually small

when compared with the benefit of added inertia.

(d) Governor Timing

Selection of the proper governor timing has the greatest overall

effect on the frequency fluctuations. It is, of course, Inter-

connected with the water hammer. Faster governor times give

better frequency control hence, lower required inertia, for

the same band width. However, faster governor times mean

higher water hammer and hence, higher inertia requirements.

Also, the maximum permissible water harrrner, which can be

tolerated by the turbine governor for stable operation is

• about 50%. It Is usual to select a governor time which will

give 50% water hammer and then determine the required inertia,
• -9-

since it is found that any increase in governor time to reduce

water hatl'll'ler has the nett effect of increasing the required

inertia for the same frequency band width, as will be illus-

trated later on.

(e) Type of Load

The type of load to which the generator will be subjected

must also be considered. One of the worst loads which can be

encountered (as far as stability and frequency control is

concerned) is that demanded by large electric shovels. The


shovels demand a heavy output over a short period of time,

particularly when the shovel operator stalls the motor. On

a large mining operation it is often possible to have two or

three shovels demanding full stalling torque at the same

instant, resulting in a sudden increase in load. Also, the

varying and pulsating nature of the load does not contribute

toward s stability.

Another type of load, is that produced by large electric

boilers of 2 or 3 megawatt capacity which can be suddenly

switched on to the system. However, frequency fluctuations

produced by boilers can usually be ironed out by having the

boiler load added in steps instead of in one unit.

• The rotating inertia of the load must also be considered •

This is usually about 10% to 25% of the generator inertia,


• -10-

and, since it is almost impossible to determine with any degree

of accuracy, it is often neglected, a procedure which will

give a value of the speed variation slightly higher than will

be encountered in practice.

6. UNIT INERTIA CHARACTERISTIC

Consideration of all the above effects has led to the development of

the chart shown in Fig. 2 on which can be plotted most of the factors

which have a bearing in t he determination of the generator inertia.

On the abscisse ls plotted the water hammer coefficient which is simply

l°/s
• where I° and

-
8 are the Allievi coefficients. and

where Lr • length o f surge tank risef r-

Ls • one-half length of scroll case

Lp - length of penstock

V • average velocity of water at full load in penstock, scroll-

case and draft tube

g = accelerat ion due to gravity


H • nett head on unit

Te • Effec tive governor opening time

• The higher this factor, the greater will be the effect of water hammer •

Normally, a value of0.4 is rarely exceeded since this will give about
• -11 -

50% water hammer above static.

In the calculation of water hammer, the length of the draft tube water

passage has been neglected, but the water velocity In the draft tube has

been included when estimating the average water velocity for t he whole

conduit. It has been found that this gives a more accurate resu lt than

inclusion of the draft tube length, justified by the fact that the

effe.cts of water hammer in t·he draft t ube are reduced to a considerable

extent by the action of the turbo vent.

On the ordinate is plotted the inertia coefficient.

2 2
• W R " N
6
I• 6 " I O x t\P ,c Tc
where WR2 a inertia

N = rpm

H a Rated HP of unit

T a Total governo r opening time


C

and is the amount of flywheel effect In the generator per horsepower per

second of total governor opening time.

In the determination of the water hammer coefficientf/e the effective

governor time Te has been used, since the maximum water hammer is produced

by the maximum rate of travel of the wicket gates.. In the inertia

• coefficie nt, the total governor time T


C
is used to allow for the slowing
• -12-

down of the rate of movement of the wicket gates under partial load

changes. Usually, the total governor time is 1.0 to 1.6 seconds greater

than the effective governor time.

The governor opening time has been selected because it ls often slower

than the closing time, a restriction imposed by either the draft tube,

or by a knee in the penstock, reducing the allowable negative water

harrmer, and hence, increasing the governor opening time.

On plotting these characteristics for over 40 hydro -electric developments

which Montreal Engineering has designed, a distinct pattern emerges. The

• development s fall into three general areas. On the left hand side fall

units which have a very low inertia per horsepower per second of

governor opening time . These units operate on base load, several having

very long penstocks with slow governor times, others having relief valves

to alleviate the effects of water hammer. They are not used to control

the frequency of the system.

In the center of the chart lie units which are connected to isolated

systems of moderate size. Generally, these units operate satisfactorily

with the standard minimum inertia when three or more units are supplying

power to the system.

On the right lie the isolated units supplying mining loads, having

generators with extra inertia. These are the units for which a careful

• analysis of the inertia is required . If the load swings are large with
• -13-

relation to the unit capacity, it is not unusual for these isolated

units to have generators with 60% to 75% extra inertia over standard.

On this basis two lines may be drawn, separating the base load units,

system units and Iso l ated units. Since the chart has no allowance for

the type of load, it can be expected that some units will not fall Into

their r espective areas. For example, three isolated units are shown

in the area pertaining to system units. The reason for this is that

two of the units will only be subjected to small load changes, and hence,

do not require an excessive amount of inertia, while the third , has


several generators of the same size, and hence, is a borderline case,

probably more accurately described as a small isolated system.

Also, several system units pl ot in the area of the isolated units

indicating that they have more than adequate inertia. Many of these

were designed as isolated units to supply a local load during failure

of the transmission line connection to the main system. Others are

units with large low head slow speed turbines, with large diameter

generators having a very high (normal) inertia, and others have hy-

draulic characteristics which automatically give good regulation with

generators of standard inertia.

This chart also illustrates why it is preferable to use fast governor

times with a large water hammer effect. Any increase in governor t i me

• will move the plotted position of a unit towards the zero, crossing the

the lines separating the areas. Thus, an isolated unit with a governor
,

• -14-

time of 3 seconds plotting in the area indicating satisfacto r y operation

as an isolated plant, would, if the governor time were increased to 10

seconds, appear in the area reserved for base load units, indicating

that control of the system frequency would not be adequate with the slow

governor time,

7. CONCLUS I ONS
Thus it is apparent that there are many factors which affect the deter-

mination of the generator inertia, ranging from the hydraulic character-

istics of the development, to the size of the system and the nature of


the load. The use of the data on Fig, 2, should, since it is based on

plants now in operation, greatly help in the selection of generator

inertia, consistent with economy and good frequency regulation,

• JLG/jh
• • • ' .

200

150
/
100 I /
/ / / /
/

/
/
90 / / / ., ./

80 / / / / ./ /

70 I / / / / /
I / / / / /
60

50
J / / / / /
40
E>:GPS
WR
ova< STMOA~
I I I / "
/ /
r--.. I
t/ ,./ . •✓ ,
"''
, ~
...,,
,-
---- 0~ ()
i---,oo
,. .
(,i
,
30

..I I I I II / /V
20
....
,.,. /
I; 1/ I;
11

. .,.
>
..
14

"
II

; 10
I I
I
I / /
; 9
8
I I I I / 5 6 T
~ I I I I J APPUCAIILE ONLY TO VERT1CAL HYOIIAUUC TURBINE
7

6
I I I / I ORIVEH SYNa<RONaJS GENERATORS,

5
I I J I ) I
4
I I I I I ,I UNIT START-1.P TIME
r, • Hx2·7 KVA

I /1
11P

/ J I
3
2 3 ..,
GENERATOR INERTIA CONSTANT H = 2·31xWfixN~ IO-IO G')
MVA ---
.
I • I ..
.

.• 0 .I .2 .4 7
·' ' lO II
" I L4 1.5 1• 17 1,

A
/.. I e " 0
,t
0
0
0 I
.,
<-
'
UNITS IN THIS RANG( CANNOT Pfl()\tlO[ AltY FA[OIJ!NCY lt<GOL ATIOW Ufft.E U.
Ft TTEO WITH RELl(F YA.I.VE$ ~RATING ON ,orH l.OAO-OF ANO LOAD-ON /
I
I fREO\lfNCY
"
" MT< I N THI$ ••••• • ..., ••
IIIEGUL TIC)N Off UltGf
ri 0 /
0
0
El
rn
..
'
t WAJUt WASTING ) ·1 SYSTEMS ONLY

i-!-I"' I
rn
I
G

>
II
"' "' 0 I I
I
.. .2
A j ~
,. 0
II I ·2
~~
j'~
+
"
..J
Q
' I I I')
I')

/
0

. 1/ I')
"' e
"'
I 1 .... ~ IN T14S RANGE Wl\.L PflOVtOE QO()Q • G'" AT...,...
IN ISOLAT£ 0 OP(ltATf()frl ~ $Y5T(M OPUAT N

a, ,.
A
J
A
I I
0
0
' .2 • • ,.o
.. I. I 1,2

0
0

,., ,
LEGEND
.
ISOLATED UNITS
SYSE:M UNITS
"
8AS£ U)AO SYSTEM UNl'l'S
1.6
'·' 1•
0

You might also like