Cooling e
Cooling e
Cooling e
In 1988 Mr. Feistel joined the R&D Group of Burckhardt Compression in Winterthur.
His responsibilities are now developing of oil-free sealing systems. In 2002 he gained
his Ph.D. degree at the University Erlangen-Nuremberg, Germany, with a thesis on
the operational behavior of dry-running sealing systems in crosshead compressors.
In 1991 he joined the Research & Development (R&D) group of Burckhardt Com-
pression in Switzerland. Since 1997 he is the head of the R&D department.
His main projects are in the field of finite element analysis, computational fluid
dynamics and mechatronic.
2
Heat Dissipation from Piston-Rod
Sealing Systems
Different concepts with consideration of the API 618
N. Feistel, G. Samland, B. Fejzuli
The need for cooling piston-rod sealing systems (packings) has recently been a
subject of more intense discussion. This has been accompanied by developments
of new cooling systems permitting more efficient heat dissipation. These activities
are motivated by a general susceptibility of commonly used cooling systems as
well as their poorer heat dissipation properties compared with cylinder cooling. An
analysis of these two shortcomings has led to a new configuration of cooling chan-
nels. Operators now have at their disposal a robust, reliable and easily maintainable
cooling system which also complies with the recommendations of the future, fifth
edition of API 618.
3
2 Designs and applications of various 2.2.1 Cooling jacket
systems for dissipating heat from piston-
rod seals A technically very simple method of heat removal by
means of a fluid is provided by a cooling jacket which
2.1 Uncooled piston-rod sealing systems surrounds all packing cups of the sealing system and
whose coolant inlets and outlets are usually directly
The basic question arising during the design of a piston- connected to the cylinder cooling (Fig. 1). The small
rod sealing system is whether or not to use liquid cool- number of gastight sealing elements over which the
ing. In the case of dry-running piston-rod sealing systems pressure difference is distributed results in a sealing
comprising non-metallic sealing elements, API 618 rec- system of widely differing load zones, the maximum
ommends liquid cooling from a pressure of 1.7 MPa values frequently occurring at the packing boundaries.
onward. Below this pressure, piston-rod seals should at One disadvantage of heat dissipation via a cooling
least provide for cooling. jacket is that its design does not ensure a directional
flow, supplying an adequate quantity of coolant to
Tests have shown that cooling might indeed be necessary zones subjected to high thermal loads. A problem simi-
below a pressure of 1.7 MPa /3/. Depending on its design, lar in principle was revealed by theoretical investiga-
a dry-running sealing system can generate critically high tions of heat dissipation from the cylinder /5/.
frictional powers especially at high average piston velo-
cities, so that cooling might be needed to ensure reliable Another disadvantage of this heat dissipation concept
operation and maximize service life. results from the additionally necessary heat transfer
from the packing cups to the cooled sleeve. Depending
However, the additional requirement for an upgrade to a on design complexity, a small gap also needs to be over-
cooling system can hinder an optimization of uncooled come here by means of free convection, which impairs
piston-rod sealing systems, their design having been heat dissipation to a correspondingly high degree. Con-
focused on a minimization of heat transfer resistance. sequently, heat dissipation via cooling jackets is
Of special importance here is a complete avoidance of restricted almost entirely to oil-lubricated piston-rod
any gaps and a use of materials possessing a high ther- seals or to dry-running packings subjected to low load
mal conductivity. At the same time, an attempt is made
to minimize friction power by optimizing sealing ele-
ment design. A key parameter in this context is the seal-
ing elements’ axial width, whose reduction results in a
corresponding decrease in friction power without any
notable deterioration in sealing efficiency /4/.
4
parameters. Advantages of this design are its simple O-rings permitting coolant to enter the sealing element
design and very low risk of coolant leakage into the chambers or gas to leak into the coolant channels must
sealing element chambers. be regarded as a disadvantage of this cooling channel
arrangement. Accordingly, the fourth edition of API 618
approves O-rings surrounding the piston-rod only from
2.2.2 Packing cup with open cooling channel a pressure of 13.5 MPa onward. The upcoming fifth edi-
tion will advise entirely against a use of circumferential
In this heat dissipation concept, the packing cups are O-rings. A confined arrangement of such rings in a
furnished with a ring-shaped cooling channel, each suc- pressure-balanced environment will continue to be per-
cessive cup closing one side of the channel (Fig. 2). The missible only for axial connecting channels (for a better
coolant is supplied by an axial bore and leaves the ring- overview, these components are not shown in the follow-
shaped channel after flowing around the piston-rod via ing illustrations).
an additional connecting channel to enter the next pack-
ing cup. In this manner, the coolant flows through all
packing cups at a constant, clearly definable flow rate, 2.2.3 Packing cup with totally enclosed cooling
but alternately in clockwise and anti-clockwise direc- channel
tions (Fig. 6).
For dry-running packings at pressures of 1.7 MPa or
One advantage of this cooling channel configuration is more and oil-lubricated designs at pressures of 3.5 MPa
its easy accessibility for the purpose of cleaning (refer or more, API 618 recommends liquid-cooled packing
to 3.2). Furthermore, the packing cups can be made cups with totally enclosed cooling channels. To improve
very narrow in the axial direction, thus permitting a the cooling efficiency, the cooling channels are usually
correspondingly large number of sealing elements to be integrated directly into the web of the packing cup,
integrated over the total length of the sealing system. right between the sealing elements in each case, in or-
Generally, the open side of the cooling channel is sealed der to minimize the heat transport distance (Fig. 3).
by means of two O-rings surrounding the piston-rod.
The latent risk of damaged, wrongly inserted or missing
Fig. 2: Piston-rod sealing system with heat dissipation by Fig. 3: Piston-rod sealing system with heat dissipation via
means of packing cups with „open“ cooling channels packing cups with totally enclosed cooling channels
5
Compared with the previously described variant involv- piston-rod via the leakage gas to the metallic packing
ing open cooling channels, however, integration into the cups, the leakage rate playing an important role here.
packing cup web can result in a larger axial dimension The limitations imposed on the cooling of a piston-rod
of the packing cups (Fig. 4), possibly necessitating at sealing system by indirect heat dissipation have already
least an elimination of a sealing element over the total been investigated in /1/ by means of a theoretical model
length of the system. The outer edge of the packing for calculating sliding surface temperatures. Variations
cups in some designs is equipped with screw plugs al- in the temperature difference between the sealing ele-
lowing the channels to be cleaned. In the case of the ments and chamber wall do not significantly influence
more widespread designs without cleaning apertures, the temperature of the friction surfaces.
the cooling channels are very awkward to maintain.
Q
Q
l ax
lax
Fig. 4: Open (left) and closed configuration of cooling channels Fig. 5: Differences in heat transport from the sealing elements’
in the packing cup and the resulting influence on the axial di- friction surfaces to the coolant: directly in the cylinder (left)
mension lax and indirectly in the piston-rod sealing system
3 Efficiency and shortcomings of cooling Consequently, measures to improve the packing cup
systems for piston-rod seals region have only a very negligible effect on the overall
heat transfer efficiency. Accelerating flow or making
3.1 Limitations of heat dissipation from a piston-rod use of turbulence mechanisms, for example, mainly
sealing system serves to just increase the friction losses of the coolant.
However, API 618 specifies a maximum permissible
In the cylinder, friction heat is transferred from the seal- coolant pressure drop through the packing case of only
ing surfaces of the piston rings to the cooling water 0.17 MPa. The original goal, i.e. a notable reduction in
directly via the cylinder wall which has a high thermal the temperature of the friction surfaces, is not achieved
conductivity. To achieve comparably good conditions with such measures. Nor were tests conducted with dif-
for the sealing elements of a piston-rod sealing system, ferent flow rates of cooling water able to reveal any fur-
the cooling channels should be placed directly into the ther drop in piston-rod temperature above a critical
piston-rod. As described previously, however, these minimum value /3/.
channels are at best integrated in the packing cups. The
heat conductivity of roughly 0.5 W/mK of plastic mate-
rials - such as a typical PTFE, filled with carbon/graph- 3.2 Shortcomings of contemporary cooling systems
ite - commonly used for sealing and rider rings reveals for piston-rod seals
that heat conductivity cannot play a major role in heat
dissipation. In contrast to the conditions prevailing in a Even before the compressor is started, liquid cooling
cylinder, heat in a piston-rod sealing system is accord- can pose considerable problems if faulty assembly en-
ingly transferred indirectly from the friction surfaces to ables coolant to penetrate into the sealing element
the cooling channels: First to the piston-rod, then by chambers. Following contact with the coolant, at least
means of convection to the metallic packing cups and dry-running sealing elements no longer guarantee
from there to the coolant (Fig. 5). proper functionality, making their use inadvisable. This
problem is compounded by elaborate cleaning required
The bottleneck forming part of the overall heat transfer in the region of the sealing element chambers and leak-
resistance from the friction surfaces to the cooling age gas piping. Particularly sensitive to this type of
channels comprises convective heat transfer from the damage are packing cups with open cooling channels if,
6
during the assembly of the packing, O-rings were either nels‘ condition is immediately apparent during servic-
left out or damaged. ing and any required cleaning can be carried out rela-
tively easily.
During operation, the gas pressure is normally higher
than the coolant pressure. Especially in the case of pack-
ing cups with open cooling channels, negligent assem- 4 The concept of longitudinal cooling
bly can therefore cause process gas to enter the cooling
system. If a closed cooling circuit has been employed, Compared with the cylinder, indirect heat transfer via
leakages of flammable, explosive or toxic gases can the piston-rod on the one hand limits the efficiency of
pose problems. Special attention needs to be paid here heat dissipation from the sealing elements’ friction sur-
to piston-rod sealing systems designed to handle high faces, and on the other hand does not allow the piston-
pressure differences in hydrogen compressors. rod temperature to be reduced significantly through
improvements to the packing cup region. Accordingly,
Especially dry-running piston-rod seals subjected to developments in piston-rod sealing systems should be
high loads depend on a perfectly functioning cooling geared primarily toward eliminating the fault suscepti-
system in spite of the latter‘s limited efficiency. How- bility of the cooling systems. One important boundary
ever, the coolant employed here is often of low quality condition here is not to fall short of the efficiency which
and purity, use sometimes even being made of perfunc- can be achieved by cooling systems commonly used
torily cleaned river water. Furthermore, operation may today. Finally, the recommendations of the future, fifth
result in a deposition of contaminations, lime etc. in the edition of API 618 need to be considered as well.
channels, thus gradually impairing heat dissipation and
ultimately leading to a thermal failure of the sealing These objectives were achieved through a new configu-
system. Compared with packing cups possessing a ration of the cooling channels in the longitudinal direc-
closed cooling channel, the design with an open cooling tion, parallel to the piston-rod, as opposed to the com-
channel here proves advantageous: The cooling chan- monly employed circular flow around each individual
Fig. 6: Conventional flow through a piston-rod sealing system (left) compared with the longitudinal channels of the new cooling
system (right)
7
The commonly employed, circular arrangement of cool- tests were conducted with hydrogen at a suction pres-
ing channels around the piston-rod encourages a depo- sure of 1.6 MPa, a discharge pressure of 4.0 MPa and
sition of solids, since the centrifugal forces occurring an average piston velocity of 3.19 m/s.
during deflection result in a separation of the coolant
from the soiling it contains. This problem is avoided by An infrared sensor was used to measure the piston-rod
longitudinal flow. Furthermore, the bores in the packing temperature during operation /3/. One problem posed
cups can be inspected very easily for soiling and cleaned by such comparative tests is that the large number of
without any trouble if necessary. The tangential channels variables influencing the piston-rod temperature makes
for diverting the flow are sealed by means of screw it very difficult to achieve similar experimental condi-
plugs which also permit easy inspection and cleaning. tions. Gas leakages through the sealing system play a
This offers operators an API 618-compliant, reliable very influential role here. Fig. 8 shows the piston-rod
and easily maintainable cooling system for piston-rod temperatures and gas leakages (measured in standard
seals. Table 1 lists the properties of the new longitudinal cubic meters per hour) ascertained for the new longitu-
cooling system compared with those of some other dinal cooling system at a cooling-water flow rate of 150
cooling systems commonly used today. l/h. These temperature values comprise averages of the
Table 1: Properties of the new longitudinal cooling system compared with those of some other cooling systems commonly used today
Leakage [scm/h]
not reduce heat dissipation efficiency. Consequently, the
new system‘s performance needed to be evaluated in 82 1.30
comparison with two widespread designs involving 80 1.25
open and enclosed cooling channels. The temperature of
the piston-rod served as a criterion for rating efficiency. 78 1.20
76 1.15
The dry-running piston-rod sealing system used for
these tests consisted of a throttle ring and five polymer- 74 1.10
blend sealing elements of the step bridge design com- 72 1.05
bined with a three-piece radially cut cover ring. The
70 1.00
sealing-ring pairs with an axial dimension of 14 mm 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70
were inserted - together with an anti-extrusion ring Time [hours]
made of modified PEEK - into chambers possessing an
axial dimension of 18 mm in all the investigated cool-
Fig. 8: Influence of gas leakage on piston-rod temperature
ing systems. The packing cups with open channels as
well as those of the longitudinal-flow cooling system
each had an axial dimension of 30 mm. However, for
the packing cups with totally enclosed cooling chan- temperature distribution along the stroke of 160 mm /3/.
nels, the integration of the channels into the stem en- Clearly evident is a drop in temperature as the leakage
tailed an increase in the axial dimension to 36 mm. The rate increases.
8
Since the leakage values were not constant during the 90
three comparative tests, a decision was made to repre- Open channels
88
sent the piston-rod temperature as a function of the Enclosed channels
leakage rate. For the three examined cooling systems, 86 Longitudinal channels
Fig. 9 shows the piston-rod temperatures measured at a
84
cooling-water flow rate of 150 l/h at intervals of 5 min-
Temperature [°C]
utes over a test period of 70 hours in each case. All 82
measured values are located within a temperature band-
width of roughly 7 K, the absolute values decreasing 80
with the leakage rate. The variations in temperature 78
clearly indicate that apart from the leakage rate, other
factors such as the running-in period and the pressure 76
distribution inside the sealing system also exert a major
74
influence on the piston-rod temperature. However, none
of the three cooling systems turned out to have any excep- 72
tional advantage. In particular, no clear advantage was 1.00 1.10 1.20 1.30 1.40 1.50 1.60
revealed by the design comprising closed cooling chan- 1.05 1.15 1.25 1.35 1.45 1.55
nels integrated directly in the packing cup web. Had the Leakage [scm/h]
overall length of the sealing system been limited to 150
mm, the large axial dimension of their packing cups Fig. 9: Piston-rod temperature as a function of gas leakage rate
would only have permitted a use of four sealing ele- for packing cups with open and enclosed cooling channels and
the new longitudinal-flow variant
ments. However, since an integration of cooling chan-
nels into the packing cup web does not improve piston-
rod temperature, the sealing system‘s service life would
be shortened in comparison with the remaining two
cooling systems.
6 Summary
The highest resistance arising during heat transfer from
the friction surfaces to the cooling channels is offered
by convective heat transfer from the piston-rod via the
leakage gas to the metallic packing cups. Measures to
improve the packing cup region accordingly have a very
negligible effect on overall heat transfer efficiency.
Therefore, development activities should be geared pri-
marily toward eliminating the fault susceptibility of
systems for cooling piston-rod seals. Important criteria
here are an avoidance of gas and coolant leakages, min-
imization of the risk of clogging and maximization of
accessibility for the purpose of inspecting and cleaning
the cooling channels.
9
Notation
PTFE polytetrafluoroethylene
PEEK polyetheretherketone
lax axial dimension of the packing cup
.
Q heat flow
References
[1] Heinrichs, K.; Strümke, M.:
Berechnung der Gleitflächentemperatur in
trockenlaufenden Kolbenstangendichtungen
Wiss. Z. Techn. Hochsch. Magdeburg 28, 1984,
Heft 4, S. 86 - 89
10
11
Milestones of our Company History
1834 Sulzer was founded 1998 First hermetically sealed, oil-free high-
1844 Engineering works Burckhardt pressure compressor package for
established in Basel hydrogen
1935 First labyrinth piston compressor (Laby®) 1999 World’s most powerful hyper compressor
1940 First Laby® for oxygen with 23’500 kW
1951 First hyper compressor for ldPE produc- 2000 Consolidation of all business activities in
tion at 1500 bar Winterthur; manufacturing facilities and
1971 First Laby® for LNG boil-off at offices in Basel closed-down
minus 160 °C 2002 Legal independence through a
1982 Merger of Sulzer and Burckhardt management buy-out, Sulzer-Burckhardt
piston compressor businesses to Engineering Works Ltd. becomes
form Sulzer-Burckhardt Engineering Burckhardt Compression AG
Works Ltd.
1982 First Laby® for IldPE and PP
Process gas compressor Compressors for a Lifetime Training and field service