Professional Documents
Culture Documents
Revamping Costs
Rambout A. Swanborn, CDS Engineering BV; Frits Koene, * Burgess Manning;
and Jan de Graauw, Delft U. of Technology
Inlet Separator
PU;S4A~~ lit
Interstage
Separator
10
/..,--.... (,)
c
Cl
~ ""'
V ".\ '0
I~
c
cyclones; this way, high separation efficiencies can be combined with the gas-flow distribution requirement through the internal, we still
small separator sizes and low separator pressure drops (low erosion). needed to ensure that the variations in the distribution would remain
Unfortunately, a more complicated construction is another conse- within the 25% overload limit tolerated by the internal. In case of
quence. Fig. 6, which shows separator efficiency as a function of load separator upgrading, we expected that this could be achieved reli-
factor, A, for a variety of cyclone-type and other separators,3 makes ably through the installation of gas-flow distribution baffles whose
the advantages of muiticyclone internals clear. size and location were to be determined by computational fluid dy-
Cyclone Type Internal Selection. It became clear that only a mul- namics (CFD) techniques.
ticyclone consisting of recycle axial cyclones would fulfill both the
size and separation-efficiency requirements that are imposed on the Ensuring Feasibility ofInstallation. We decided that a mockup of
retrofit internals for the overloaded separators. In a joint research the vessel as it would be after the removal of the current internal
project, carried out at the U. of Delft and sponsored primarily by would be necessary to ensure that the new internal could be located
Shell IntI. Petroleum Mij. B V and the Dutch Ministry of Economic and fastened where foreseen in case a separator-upgrade solution
Affairs, a compact recycle axial cyclone was developed. 4 This cy- was preferred.
clone had been operationally proved in on- and offshore gas-clean-
ing applications and was commercially available at the time of this New Separators or Upgrading of Current Separators?
project (the CDS Engineering A-X Axiflow cyclone licensed to
The costs of two new sets of retrofit internals (including installation)
Burgess Manning Europe). With the practical experience gained
with this cyclone, designing a reliable cyclone internal with a sur-
amounted to = $170,000. Removal of the current internals and
installation of the new multicyclones was expected to take place in
face area of 2.9 m 2, containing 600 5.08-cm cyclones and still offer-
mid-1994. These facts compared very favorably with those linked
ing a tolerance to 25% local overloading was feasible.
to the installation of new, bigger separators (Table 1). Given that the
As an illustration of the increased capacity of the separators with
risks, as previously described, for installing retrofit internals were
this internal, the maximum inlet velocity of the cyclones was = 60
well quantified and considered acceptable, we decided to proceed
m1s (compared with 25 m1s for old cyclones), resulting in G forces
between 5000 and 6000 G (compared with between 300 and 500 G). with the upgrading of the current separators.
The net free area of this internal is 48% (compared with 28%).
Ensuring Proper Flow Distribution
Ensuring Even Flow Distribution. With the availability of the de- A CFD package (Fluent) was used to determine the distribution of
scribed multiaxial-cyclone-type internal, the first requirement for a the gas flow through the internal. With such software tools, gas-flow
successful separator upgrading had been fulfilled. With respect to patterns can be predicted in various geometries.
With the cyclone internal mounted in the vessel without any fur-
ther precautions, the flow pattern schematically depicted in Fig. 7A
was predicted. It became clear that hardly any flow at all went
through the upstream cyclones, even in the wrong direction, while
the downstream cyclones handled the bulk of the flow. These cy- Fig. 7A-Predicted flow pattern through retrofit internal without
clones handled =35% more than the average design flow rate, baffles.
which is more than the tolerated overload. This flow pattern is
caused by the fact that the static pressure below the upstream cy-
clones is low (high local gas velocities)and high above these cy-
clones (low local velocities). This effect is made even worse by the
LJLq~
narrowing of the gas flow when it enters the narrow space between
the internal and the vessel.
We simulated various gas-flow patterns with different flow-baffle
configurations. The baffle configuration in Fig. 7B gives the opti- , ,
mum simulated flow pattern. The upstream baffle compensates for l ;
the narrowing effect, and the downstream baffles improve flow dis- i r-
i f
tribution through the cyclones by influencing the static-pressure ..-?/
===:...J..j.:;;;;
distribution below the cyclones. Fig. 78 shows the resulting flow
pattern: local overloading is reduced to < 15%, which the cyclones
should be capable of handling easily. Another aspect that became
Fig. 7B-Predicted flow pattern through retrofit internal with
clear with the simulations was that the installed louvre-type inlet de-
baffles.
vice had an adverse effect on gas-flow distribution under the new
operating conditions. Removing it proved to be the best solution.