HYDRAULIC INSTITUTE STA

Questions relating to proper reducer application in centrifugal pump suction lines date back
many decades. Until his death (at age 84, in 1995, !orld"reno!ned pump e#pert $gor %arassik
fre&uently corresponded !ith the !riter and other pump users on pump"related sub'ects. e
rarely pass up an opportunity to highlight some of his e#perience"based comments.

)nce, a pump user referred to *ig. 1 and noted that this !as &uite typical of illustrations found in
many te#tbooks. $n essence, *ig. 1 indicates that, !ith a suction line entering the pump in the
hori+ontal plane, the eccentric reducer is placed !ith the flat at the top. -ailable te#ts often gi-e
no indication as to !hether the pumpage came from abo-e or belo! the pump.

Fig. 1. $llustration of eccentric reducer mounting from

$gor %arassik agreed that, if the supply source !as from abo-e the pump, the eccentric reducer
should be installed !ith the flat (hori+ontal surface at the bottom. 0ntrained -apor bubbles
could then migrate back into the source instead of staying near the pump suction. $f the pump
suction piping entered after a long hori+ontal run or from belo! the pump, the flat of the
eccentric reducer should be at the top.1

/till, in many older te#ts it has been assumed that the pumpage source originated at a le-el
belo! the pump suction no++le. %arassik reminded us that older ydraulic $nstitute /tandards
commented on the suction pipe slope

2...ny high point in the suction pipe !ill become filled !ith air and thus pre-ent proper operation
of the pump.  straight taper reducer should not be used in a hori+ontal suction line as an air
pocket is formed in the top of the reducer and the pipe. n eccentric reducer should be used
instead.3

his instruction applies regardless of !here the pumpage originates. epending on the
particulars of an installation, trapped -apors can reduce the effecti-e suction line cross"sectional
area. /hould that be the case, flo! -elocities !ould tend to be higher than anticipated. igher
friction losses !ould occur and pump performance !ould be ad-ersely affected.

$n the case of a li&uid source abo-e the pump suction, and particularly !here the suction line
consists of an eccentric reducer follo!ed by an elbo! turned -ertically up!ard and a -ertical
pipe length6all assembled in that se&uence from the pump suction flange upstream6it !ill be
mandatory for the eccentric reducer flat side to be at the bottom. hat said, *ig. 7 should clarify
!hat reliability"focused users need to implement.
 

HYDRAULIC INSTITUTE STA

Fig. 2. /uggested modifications for eccentric reducer 

 lso, !hene-er
!hene-er -apors
-apors must be -ented against
against the
the flo! direction,
direction, the line si+e upstream
upstream of
of any
lo! point must be go-erned by an important criterion. he line must be a diameter that !ill limit
the pumpage -elocity to -alues belo! those !here bubbles !ill rise through the li&uid.

$n general, it can be stated that !here-er a lo! point e#ists in a suction line, the hori+ontal
piping run at that point should be kept as short as possible. $n a proper installation, the reducer
flange !ill thus be located at the pump suction no++le and there is usually no straight piping
bet!een reducer outlet   and
and pump no++le. /traight pipe lengths are, ho!e-er, connected to the
eccentric reducer inlet  flange.
 flange. )n most pumps, one usually gets a!ay !ith fi-e diameters of
straight length ne#t to the reducer. $n the case of certain unspecified -elocities and other
interacting -ariables (e.g., -iscosity, / margin, pump style, etc., it might be !ise to install
as many as 1: diameters of straight length ne#t to the reducer inlet flange. he t!o different
rules"of"thumb e#plain seeming inconsistencies in the literature, !here both the 5 and 1:"
rules can be found.