Information and Recommendations for the Engineer
FAN ENGINEERING #=@y=w
Surge, Stall, and Instabilities in Fans
Introduction Figure 2. Typical Flow
Usots of fan systoms desi a steady, continuous flow v
of ait. In this idoal situation, the pressure generated by
‘the fan is constant. A single’ instantaneous measurement
of the flow rate would De valid for extended periods of =
ime. 5
Figure 1 shows the flow for an ideal system. Figures B@
2 through 5 show a variety of conditions for time varia- -
ton of flow.
Figure 7. Ideal Flow
Figure 3. Stal Condition
Bw
‘Those involved with measurement of flow rales know =
that ideal flow conditions are not common, Each point :
of flow measurement is usually time averaged for ten p reer
seconds of mofo fo got an accurate reading. Vatations rate
In flow and prossure readings of 10% over short time
porlods aro relativoly common. Howover, fans which are Figure 4. Suige Condition
Improperly selacted oF applied can produce variations fat
greater than this. Conditions can become so severe that
‘he flow through the tan can oscllate between forward
ana reverse (Tow exiting the inlet) many times per min-
Ute (see Figure 4).
‘The variations in tiow and pressure not only make st
more diffcut to measure the flow, they can create a
variaty of problems:
1. Dramatic Increases in noise.
2. increases in vibration,
3 Stuctural fatigue damage to the tan due to continu-
‘ous loading and unloading of components.
4. Damage to the ductwork and other system compo-
nents
5. A fan system that does not perform properly due to Figur 6. B-Siable Fow Gondiion
unsteady Now and/or transmited vibration a
‘An understanaing of the causes of unsteady tow can 0
be helpful m avoiding these problems. Because sore ot I
‘the causes. are very complex, researchers nave nad ~
some interest. However, thete nas not been uniform
agreement in the conclusions as to what the exact
causes are. From thelr research we can learn the cond- =
‘ons that tend to perform normally and avoid the cond
tions that don't
Systems with unsteady flow can perform mysteri- .
ously. Complicated terms to describe the phenomena
are Often used and misused. The result is that there are
(22012 Twn Oiy Fan Gempanis, Lsfow authoritative reference materials to use as diagnos-
tic guides. We will discuss some of the more common
loris and Issues.
Stall
Elongated objects (such as fan blades) when passing
‘through the airstream will deflect the air. I we change
the ofiantation of the objact relative to the flow direction,
we can increase or decrease the amount that the ait
deflected. If me progressively increase the attack angle
Of iho fan blade, it wil increase the amount of dortec-
tion. TL Is this chango of direction (and relative vetocttcs)
that allows the fan to gonorato pressure. If tho attack
anglo becomes {00 savor, the ait wil NO longer follow
‘he lage surface In @ uniform manner. The amount ot
Geection and tne pressure Deng generated stops
Increasing and normally wil tall off. This ‘6 called the
stall point. Figure 6 shows a fan curve of a fan with @
significant dip In the stall region.
na fan, the blades are normally rotating at constant
velocity. Therefore, to change the angle of attack, the
system to which the fan is attached must bs changed.
Higher flow rates through the Inlet increase the flack
angle, and lower flow rates decrease it. Therefore, it a
fan IS operating in stall, t Is becauso the CFM Is too
low for i. On a gon systom, this ' caused by solact-
Ing a fan which Is too large’ (making the air voioctties
{09 Tow in tho fay,
Figure 6. Curve Showing 2 Dip in the Stal! Region
4
5 pu
NS
ase
STATIC PRESSURE (inWWC)
Lu
reereeeberrenenterrnet ei
o 4 8 Re Oe
‘CFM in (1.0005)
In some fans, the angle of attack is not uniform
across the full width of the blade. These are normally
fot the most efficient fans, although the severity of the
stalls often less since only part of the blade is staling
al any one flow rele, Some people say that radial
Bladed contitugal fans aro alwaye in stall sinco thoro fs
poor “match botwoon tno droctonal velocity of tho
Diade and that of te approaching arr. Ths 1S essen-
‘aly trus. However, these types of fans can have severe
varying Tiows at very low Tow rates since the intemal
losses are dominated by stal and the pressure fais of
at ths pont.
Avian operating at or near the stall point usualy wil
have severe increases in noise. On some fans It Wi
sound almost as if the ipaller is being impacted by 2
sold. object (hammering) Pure stall fends to have. &
random frequency but there are special cases where &
pure ffequency Is generated. Ths wil be discussed
late.
‘There is 2 time-varying nature for the flow of a fan
in stall. However, this fs normaly not the major cause
fof concern, The increased noise being generated can be
4 probiom, but this too can often bo dealt with. Tho
Major concern for a tan operating in stall is the poten.
Lal for mechanical damage. Thoso who havo had a
“pumpy” alfpiane ride nave & feel for now severe aero-
‘ynamic shook impacts can be.
‘A. an Continuously Operating install can_sustan
structural metal fatigue. This Is especialy tue tor axa
‘low fans having long slender blades, or blades fabr-
cated from snest metal. Centrfugal fans are less. prone
to damage. Centritugal fans designed for relatively high
pressures but operating at very low pressures (ess than
4" SP) have been known to operate continuously in stall
for many years without damage
“Thora Is another downside To having a tan oporate in
stall It moans the officioncy of the fan Is less than
‘optimum. A smaller sizo fan costs loss and has a lowor
‘oporaling cost. It wil also IKaly outlast a larger tan,
Rotating Stall
This IS a special case of stal that normally only occurs,
In backwaraly inclined ang aioli centritugal fans. Most
observers also. report that inlet box dampers are
involved. Variable inlet vanes do a good job of prevent-
Ing rotating stall Decause they provide a more stable
‘low path for the air through the wheel. These fans are
fencased in a scroll type housing that helps generate the
fan's pressure. The pressure around the periphery of the
fan wheel varies relative to how near it Is to the fan
‘outlet (Where it is highest). These fans have several
blades, typically 9 to12.
‘We’ will call the passageway between each blade 2
“call” Tho flow through dach coll can vary sinco tho
pprossure around the periphery vaties. Near the stall point
TR becomes possible for most of tne cells to nave the
Normal forward Tlow, while one oF {Wo cells nave reverse
‘low. The air that “squirts” backward through these cells
has nowhere to go so it moves to an adjacent cel
eflecting the air which was already traveling through ft
This change of attack angle now forces this cell to stall
it then also has reverse flow, passes on fs bubble of
air, and-on and on around the’ fan wheel
‘Rotating stall typically occurs in fans which are
severely throttled (Inlet box damper typically less than
30% oper).
Most researchers have reported that the frequency of,
travol of this rotating stall occurs at about two-thirds of
tho fan rotational RPMO9. Some have observed. two
traveling cols al once generating a four-thirds rotational
frequoncy. There aro other reports of rotating stall rang
Ing from two-tniras and even higher harmonics (2/3.
4X, G/3K, B/3K, ..). It these excitng Meguencles coin”
‘olde’ win the naturai wequencies of the wneel or nous-
Ing, resonance cecurs and damage can resuit. This
frequency will show up in both sound and. vibration
measurements. Rotating stall is among the most destruc-
tive of instabilties in the fan.
Surge
Several yeas ago there was a report of a gan dying
Syotom at was proscar lage gran Ore 1 llaed
’bumor socton near to hit a ergo conta fan
Ponodaly Ine sjtam woud “Dot ro Dac out
Ina of to burns The was IY a" Sovae ato of
system sue.
"ne coun a upg. Ton makes. Ie. commony
cecorbad by eooerare ie ungost or “anoome, Tete
fe cowl ottena Wich mt be met To nave sue
Fan Enginooring FE-6004. There must be a relatively large volume of air which
Js pressurized (such as the grain bins or a large ple-
num).
2. Thord must be a soction of ductwork with relatvoly
high voiocitis.
3. The operating point of tho system Is to the lett of
the peak pressure (at lower low rates). In tnls region
the fan curve has 2 positive slope such that Inoreas-
Ing te flow also iereases the Tan static pressure.
In concept, a system in surge Is tke an oscillator.
‘The enoray imparted to the air alternates botwoon croat-
Ing Kinotc energy igh voloctty in the duct) and poton-
tial onergy (compressing tho air in tho plonum). The
[posite siops on the fan curve allows large amplification
Of this osollation to occur. in extreme conditions, the ar
‘can ‘emporanty biow back trough the inet.
Ina txed system, the frequency ot the surge Is con-
stant, Usually the ftequency Is low enough that you can
count the number of cycles per minute since it Is quite
aucible. Mast savere reports secur at a frequency below
300 cpm. One researcher reported that this effact seems
to disappear al frequencies above 450 cpm.
The frequency of sue can be be calculated for
simple systems:
Frequency (Hz) = 175 + Square Root [Duct-area /
(Plenum-volume = Duct-length)]
Note: Koop all dimensions in foot,
For those wishing further information, research the
teem “helmmoltz resonator.”
Hunting
Some people use the tem “hunting” to apply to any
time-varying flow. However, tho appropriato uso of this
farm should apply to an under damped control circu.
In variable volume systems, sensors are Used to provide
Information that controis dampers, vanes. speed contro’,
or other means of setting the flow rate. If the control
system responds too quckly, ft wil overcorrect and have
10 readuust the otner diection. In the extreme condition,
fa system may continuously “hunt” back and forth
Stability and Bi-Stable Flow
Stablity refers to the ability of a system if temporarily
displaced to tetum fo ifs Un-displaced position. coin
balanced on edge is unstable. A coin at rast on a flat
surface is stable. Some fans are not slable for all flow
ranges. For example, walking by the inlet (don't try thish
of a large conttfugal_during an air test caused tne flow
{0 foduce by over 15%. This fan continued to oporate
al tho lower flow rato until the tost was restarted.
We can detormine the stabilty of a fan by parforming
two alr tests. On one test, we star at rll Tiow (ree
aelver) and measure tne flow and pressure as we
rogresswely ada_more resistance, In the second test,
we start st shut-off and progressively reduce the resis:
tance. We now have two flow vs. pressure fan curves,
If they do not overlay, we have a region of instability
Figure 7 shows 2 sample fan curve with this properly
Since there are only two possible conditions on any
system, this iS called b-stable flow.
‘Although the noise changes batwesn the two flow
conditions, neither is particularly objectionable. if tho tan
's ralod In tho high flow condition and it “trios” to the
lower condition, tne loss of flow can be 2 problem,
Bi-stable flow nas bocn obsorved in backwardly
Incined centrfugal tans, usually at pertormances close
10 tee delvery and amost aways at low rates nigner
‘han that wnere the best efticiency occurs.
Figure 7. Bi-Stable Flow
of \
STATIC PRESSURE (In¥/C)
crm
Parallel Flow Operation
Its relatwely common tor two of more tans to be oper-
ating in parallel In the two fan system, each tan is
selected for hait of the design flow rate. Fans that nave
a large “sip” in the stall region can have another type
‘of problem. Vaneaxial and forwarc) curve centrifugal fans
are_two types of fans which can have large “dips.”
“The problem with parallel flow systems can occur in
the stafting sequence. If the fans are properly sized
started simuttaneously and brought up to speed at the
same rale, there is no problem. However, if one fan is
sailed fii, the second fan is already exposed to back
pressure while itis coming up to speed. At full speed,
2 condition can ariso whore one fan Is oporating at 2
flow rate to the right of tie peak static pressure point,
mile the otfer fan 1s tapped on the left side of the
peak.
‘Skeptics will not think this 's possible, so here is an
‘example. A fan was to be selected for 7000 CFM at 4°
SP. Due to severe space limitations, two 18% diameter
vaneaxials were selacted to operate in parallel. Referring
to Figure (8), we can see that each fan was 1
deliver 3500 CFM.
Figure 8, Two identical Fans Not Sharing tne Load Equally
3 J
g
2 z x
24
k
Fan Enginooring FE-600Those familar with the system will recognize the
system curve as 2 parabola with the equation
SP = CFNP = Constant
For this systom wo got:
4 = 7000 + Constant
Constant = 49,000,000 = 4 = 12,250,000
We can determine what this system will do with one
{an running by plotting this parabola on our fan curve.
with fan #1 running we can see that we will get anout
4750 CFM at 1.8" SP (A)
‘At ths point, flow would be going backwards through
Yan #2 unless we stop the leak. Commanly, heavy duty
backdraft dampers are used for this. When starting tan
432, It will contribute no flow until I reachos @ spood
such that the prossure (@l_shut-of exceeds tho back
Dprossuro It soos from fan #1. AS tho spood continues
to increase, tne low wil eventualy reach point. “D”.
Meanwhile, te pressure fan #1 Sees continues to
Increase unti reaches point “C”.
‘We then nave the tinal conattion:
Fan #1 Is delivering 3800 CFM @ 3.60" SP
Fan #2 Is delvenng 2750 CFM @ 3.50" SP
From the system curve equatior
(2780 + 3800 + 12,250,000 ~ 3.50"
Figure 9. Norma! watn Fan PPM=1282, BHP-13.14
es [econfirar fase dain
+ The system Is happy because the flow (total of both
Tans) and pressure readings are on the system
curve.
+ Fan #1 Is nappy because point °C” Is on the fan
uve.
+ Fan #2 has mixed emotions. It is happy that point
“G" ts on the Tan curve, Dut unhappy about being
‘wapped in stall
I these were your tans you woud probably be
Unhappy becouse:
41. The total CFM is only 6550, not 7000 as expected.
2. Fan #2 is probably noksy.
8. Fan #2 18 prone to damage due to operation in
stall
This example shows that I 1S possible to nave two
Identical tans not sharing the load equally. A more
severe condition can exist It non-identical fans are oper-
ating In parallel, Some years ago a complaint regarding
2 system with two fans in parallel was recalved trom a
‘customer. After instaling a ‘second larger fan In parallel
with a smaller fan that had Ge0n In operation, the com=
bined flow wasn't what was expected. Measurements
revealed that the second fan by iiself was. generating
more pressure than the first fan was capable of at any
point on ts fan curve. Tho original fan was. complotoly
‘overpowered, and flow was blowing back out of Its ini.
Tho customer was advised to shut off tho orginal ran
(saving the power) and block solid the duct branch to
Figure 10. Reduced Wat Fan RPM~1342, BHP-12.56
se
wd [ste
A
SC ES
BRAKE HORSEPOWER
[STATIC PRESSURE (in¥WC)
STATIC PRESSURE (in)
if
~
t
T
[BRAKE HORSEPOWER,
(CFM in (7.0008)
Figure 11. Blade Angio~a5*
Figure 12, Blade Angle-35°
Ne
4 NS
BRAKE HORSEPOWER
‘CFM in (1.0005)
BRAKE HORSEPOWER
[STATIC PRESSURE (in)
(CFM in (1.0008)
Fan Enginooring FE-600