Originally appeared in:

December 2018, pgs 51-55.

Maintenance
Used with permission.
and Reliability
E. JOHNSON, Graham Corp., Covington, Louisiana

Maintenance and turnaround planning critical

to successful ejector vacuum system operation
Steam ejector vacuum systems are generally considered a contributing factor to sealing strip damage and degradation.
to be reliable and require little maintenance. However, when Because of this, it is recommended that they are replaced each
preventive or corrective maintenance is required, ensuring it time the bundle is pulled to ensure a good seal is maintained.
is done correctly is imperative. Failing to properly perform The refiner’s vacuum distillation system undergoes mainte-
maintenance or turnaround work can result in a loss of per- nance every 5 yr during each major turnaround. The ejector
formance or worse, a loss of production capacity. Many of the system is run continuously between these maintenance outages
services that ejector systems operate in are critical services to with little to no opportunity to service the equipment. It was
the overall process. decided that the system’s condenser bundles would be pulled
Opportunities for maintenance work are also often heav- and cleaned during the turnaround. The condensers were be-
ily dependent on turnaround and shutdown schedules. This lieved to be fouled, based on the declining performance of the
further increases the importance of getting the maintenance system during the hotter months of the year. The vacuum tow-
work done correctly every time. Unfortunately, this lesson is er historically ran close to its design operating pressure. In the
often learned the hard way with improper maintenance work year or so leading up to the turnaround, the vacuum system be-
resulting in poor performance and lost capacity. Multiple case came unstable at cooling water temperatures nearing the design
studies have been outlined and selected to cover a variety of point for the system, causing degradation in tower pressure.
services and industries, and to provide useful insight so these When the ejector system was brought down for mainte-
issues can successfully be avoided. nance during the turnaround, the fouling buildup on both the
Highlighted below are four such case studies, which have tube and shell sides of the system’s condensers was removed.
been selected from a pool of examples and represent real is- The tubes were lanced to remove scaling, and the shell side
sues that have been faced by plants of various industries. The was water blasted to remove the process buildup. The sys-
case study examples highlight the thought processes that lead tem was reassembled and brought back online after the turn-
to these errors and detail the impact that these errors had on around activities.
the vacuum system’s operation.

Case study 1: The importance of sealing strips. This

case study focuses on a vacuum distillation ejector system for a
Middle Eastern refinery.
Condensers can be supplied in a variety of different con-
figurations, and fixed tubesheet designs are one of the most
basic arrangements. In a fixed tubesheet design, the tubesheets
are welded to the shell. While cheaper in cost, this prohibits
the mechanical cleaning of the shell side of a condenser. In a
dirty process, such as vacuum distillation, it is more common
for a removable bundle to be selected. Floating head or U-tube
bundle designs often utilize sealing strips. The internal baffles
cannot be welded to the shell, as that would prevent the bundle
from being removed. However, the baffling needs to seal to the
shell to prevent vapor bypassing. This is where sealing strips
are utilized, as shown in FIG. 1.
Many maintenance activities, such as cleaning, require the
bundle to be pulled from the shell. When this is done, the seal- FIG. 1. Sealing strips are utilized to seal to the shell to prevent
vapor bypassing.
ing strips can become damaged. Over time, corrosion is often
Hydrocarbon Processing | DECEMBER 2018
 Maintenance and Reliability

To everyone’s surprise, the system performance after be- The refiner procured the required sealing strips and placed
ing brought back online was substantially worse than prior to them into stock. Within 1 yr, an opportunity presented itself
the shutdown. The tower vacuum was recorded to be near 60 when the boiler that supplies steam to the ejector system had
mmHgA, when it had previously been operating at a pressure an unplanned outage. The refiner used this opportunity to pull
closer to the design point of 15 mmHgA. It was expected that the condenser bundles and confirmed that the sealing strips
cleaning and maintenance would improve the system’s perfor- were either missing or in very poor condition (FIG. 3).
mance, but the results were the opposite. This was a terrifying After the bundles were reinstalled with new sealing strips
prospect, since the system was not scheduled for additional and the system was brought back online, the vacuum returned
maintenance until the next turnaround in 5 yr. to the design of 15 mmHgA. To prevent this from reoccurring,
Discussion and onsite assistance identified that the con- this refiner has placed a hold point for engineering to inspect the
denser bundle sealing strips were not replaced during the cleaned bundles and verify new sealing strips have been installed
turnaround. The condenser bundles had either been installed prior to bundle reinstallation on all future condenser work.
without sealing strips or the original sealing strips were severe-
ly damaged. Thermal images of the condensers (FIG. 2) were Case study 2: Equipment replaced with improperly de-
taken, and the temperature profile indicated bypassing through signed copy. This case study examines a compressor drive
the sealing strips. This bypassing steam acts as a load to the steam turbine exhaust condenser at a Caribbean fertilizer
downstream and, in this case, that bypass was severe enough company.
that it was overloading the ejectors. This was directly respon- Steam turbines often utilize a condenser on their discharge,
sible for the tower’s poor performance. which services two main purposes. One reason for placing a
The economic impact of running a vacuum tower at 60 condenser at the discharge of a steam turbine is to improve its
mmHgA vs. the design of 15 mmHgA for 5 yr is unacceptable. efficiency. The condenser allows the turbine to discharge to a
vacuum, which creates a larger pressure differential across the
turbine, increasing its efficiency. The other purpose of a con-
denser is to capture the steam in the form of condensate so it
can be recycled back to the boiler. For condensers to perform
both tasks, they are designed in a very specific way that differs
significantly from similar looking heat exchangers.
This fertilizer company installed a new steam turbine ex-
haust condenser built by a third-party heat exchanger manu-
facturer. The equipment was purchased to replace an original
equipment manufacturer (OEM)-built 1965 condenser. After
more than 40 yr of service, the mechanical integrity of the orig-
inal condenser was questionable. The decision was made that
it would be more cost-effective to replace the unit rather than
trying to rebuild it, which is normally the better choice.
The equipment user sourced the replacement condenser
from a well-known and trusted heat exchanger company due to
price and lead time concerns with the OEM’s offer. The plant
FIG. 2. Thermal images of the condensers were taken and indicated in question has a 4-yr turnaround cycle, allowing for limited
bypassing through the sealing strips. opportunities to service equipment between outages. The re-
placement condenser was installed during one of these outages.
The design vacuum for the condenser is 3.75 inHgA at full
load. Upon completion of the outage, the steam turbine was
brought back online with poor results. The condenser was only
able to reach an operating pressure of 20 inHgA at 90% of the
design steam load. Although the cooling water temperature
and flow were also found to be better than the design, the unit
was nearly 16 inHg off its ideal operating point. This proved to
be catastrophic for the plant, which was steam limited. Expan-
sions to the plant left no additional boiler capacity. This meant
it could not overcome the poor turbine back pressure with ad-
ditional steam, ultimately preventing the plant from running at
full capacity. It is one thing to operate a unit inefficiently, but
it is a more serious issue when it limits a plant’s production.
The estimated monetary costs of this performance shortcom-
ing were truly astronomical and career ending.
The heat exchanger manufacturer had no field assistance
or experience with turbine condensing systems and could not
FIG. 3. Damaged sealing strips. provide the end user with support. To determine how this is-
DECEMBER 2018 | HydrocarbonProcessing.com
 Maintenance and Reliability

sue originated, the original condenser’s manufacturer was con- (4.4°C). Mounted on the top of the SVRU is a direct contact
tacted to review the problem and inspect the equipment onsite. barometric condenser and a small two-stage air venting sys-
A review of the replacement condenser drawings revealed tem. The air venting system sets the discharge pressure for two
ignorance on the part of the heat exchanger manufacturer and large booster ejectors that are pulling the deep vacuum on the
showed that the unit was incorrectly designed. It was built like flash tanks. After nearly 45 yr of operation, the SVR system’s
a heat exchanger, as opposed to a turbine exhaust condenser. capacity suddenly dropped to an unacceptable level, setting off
Although these appear similar on the outside, the internals are a series of costly errors.
very different, and they have two very different functions. It Since chilled water is critical to making paper, a mechani-
is important to remember that although condensers exchange cal chiller was rented to provide the mill’s chilled water. This
heat, their purpose is to facilitate a phase change in the process chiller would remain in service until the issue with the SVRU
flow, turning vapors into liquids. Heat exchangers are designed could be identified and corrected. The rented mechanical
to transfer heat from one medium to another. chiller was costly, prompting the user to resolve the SVR prob-
The incorrectly designed and installed unit was missing lems quickly.
most of the critical features that can be expected in a surface
condenser. It also incorporated many heat exchanger features
that are counterproductive in a vacuum condenser. The site’s
engineering staff believed that because the replacement had
the required surface area, it should have performed well. The
underperforming exchanger was modeled with heat exchanger
software that supported the poor design. Heat exchanger soft-
ware does a poor job of modeling condensers.
One major issue was a complete lack of understanding as to
why condensers are installed with an air removal system. The
air removal system is installed on a condenser to pull a deep
enough vacuum that any air leaking into the condenser can be
swept from the bundle. Air in a condenser will blanket tubes
displacing steam, resulting in poor operating pressure. The in- FIG. 4. An incorrectly reverse engineered air removal system connection.
ternals of a condenser are set up to direct the air that is leak-
ing into a vacuum system toward the air removal equipment.
In this instance, the replacement unit was pulling the non- The equipment user identified that the large booster ejec-
condensables out of a connection that was located adjacent to tor had developed a significant hole in the side of one of the
the steam inlet (FIG. 4). Most of what the venting package was diffusers, causing the issue. This was allowing air to be pulled
pulling over was condensable steam, rather than noncondens- into the system, overloading the air venting ejectors and caus-
ables. This allowed for large dead areas in the condenser where ing a significant loss in performance. Due to price and lead
noncondensables were gathering, preventing a deeper vacuum time concerns, the user had a local fabricator reverse engineer
from being achieved. both booster diffusers. The newly fabricated replacement dif-
An operational fix was unavailable for this problem; the new fusers were then installed, which successfully eliminated the
unit needed to be completely replaced. Plans were put in place air leak but never returned the performance of the system.
to replace the condenser at the next outage in 4 yr. The cost of This prevented the rented mechanical chiller from being re-
a new, properly design condenser as well as the cost of the im- moved from service.
properly designed unit was high, but ultimately that price was The equipment user, baffled by the ongoing problem, con-
small compared to the cost of the lost production. tacted the SVRU’s manufacturer to perform a site survey of
the vacuum equipment and determine the cause of the con-
Case study 3: Incorrectly fabricated non-OEM parts. tinuing issues. While several smaller issues were identified and
This case study details the poor performance of a steam vacu- corrected, the source of the continued poor performance was
um refrigeration ejector system at a US paper mill. determined to be the replacement diffusers, which were im-
An interesting application for steam-powered ejectors is a properly designed and failed to meet the system’s needs.
steam vacuum refrigeration unit (SVRU). Although it seems Examining these diffusers highlighted several glaring is-
counterintuitive, steam can be used with an ejector system to sues, including:
generate chilled water. This is achieved by pulling deep vacu- • The replacement diffusers were not concentric.
um on a water stream and flashing off some of that water, which As is often an issue with non-OEM diffusers,
in turn chills that water. Achieving a chilled water temperature the replacement diffusers were oblong, which
of 40°F (4.4°C), as the unit was designed to do, is not uncom- was negatively impacting performance.
mon with one of these systems. This is a particularly attractive • The alignment was incorrect, meaning the new diffuser
process for paper mills, since their process requires chilled wa- was not straight relative to the air chamber and motive
ter and they normally have an abundance of available steam, nozzle. When the motive nozzle is incorrectly aligned
although SVRUs are also found in other industries. with the diffuser, the motive steam will be focused
This SVRU is 49-ft tall and utilizes two flash chambers towards one side of the ejector, reducing its efficiency.
to achieve its designed chilled water temperature of 40°F Replacement booster ejectors (FIG. 5) were ordered from
Hydrocarbon Processing | DECEMBER 2018 
 Maintenance and Reliability

the SVRU’s OEM. Following installation, the performance of cess stream, making the downstream equipment smaller and
the system returned to design. Although ejectors appear simple more efficient. Condensers are designed with low process side
from the outside, precision and alignment are critical. Small, pressure drops to maximize efficiency, allowing the upstream
ejector to be designed for a lower discharge pressure
and lowering its required compression range. This
reduces the system’s steam consumption. The con-
Failing to properly perform maintenance densers and their design are an integral component
or turnaround work on critical ejector of any condensing vacuum system.
The refiner in question was operating a three-
systems can result in a loss of performance stage ejector system that was pulling vacuum on a
or worse, a loss of production capacity. distillation column. The system contained three
condensers, one between each stage of ejectors and
an additional condenser after the ejectors. Over the
course of roughly 20 yr of continual operation, the
seemingly unnoticeable deviations can be detrimental to suc- condenser tube bundles began to suffer from corrosion and
cessful performance. Going forward, this site plans a yearly wear. This specific refinery undergoes a turnaround every 2
inspection so potential system issues can be identified before yr, and it was decided that the tube bundles for the condens-
they fail, and necessary parts are stocked so that any failure can ers would be replaced during the next turnaround. Since the
quickly be addressed. condensers for this system are of a U-tube design, the bundles
can be changed out without having to replace the shell.
Case study 4: Condenser tube bundles improperly re- The replacement bundles were purchased from a reputable
placed by a third-party. This case study examines a US re- local heat exchanger manufacturer to acquire them inexpen-
finery vacuum distillation ejector system. sively. However, two critical mistakes were made by the refin-
Vacuum distillation ejector systems are normally com- ery that led to additional costs and lost production. Had the
prised of three stages of ejectors, although other arrangements bundles been replaced in kind and by the original vacuum sys-
are used. To improve the efficiency of a vacuum system, con- tem’s manufacturer, these problems would have been avoided,
densers are utilized between the ejectors. This allows for steam saving the plant a substantial amount of money and time in
and condensable hydrocarbons to be removed from the pro- the long run.
The material of the replacement bundles was changed. The
original condensers were supplied with 304SS tubes, but the
replacement carbon-steel tube bundles were procured from
the heat exchanger manufacturer. While carbon-steel tubes
are used successfully in various refinery applications, they are
a notoriously bad materials selection for a vacuum system.
The refinery selected carbon steel for two reasons: it has bet-
ter heat transfer properties than stainless steel and is signifi-
cantly cheaper. In a vacuum distillation service, the tubes are
exposed to both oxygenated cooling water on one side and
corrosive vapors, which contain trace amounts of oxygen, on
the process side.

FIG. 5. OEM replacement ejectors on an SVRU. FIG. 6. A maintenance yard filled with failed bundles.

DECEMBER 2018 | HydrocarbonProcessing.com
 Maintenance and Reliability

The original stainless-steel bundles lasted roughly 20 yr, cept the poor pressure and constant replacements as inevitable,
whereas the replacement bundles lasted 18 mos. This short as opposed to correctable. To this day, the plant in question
service life was due to the carbon steel oxidizing and develop- continues to operate under these inefficient conditions, despite
ing a multitude of cooling water leaks. Another unfortunate the evidence presented regarding the cause of these issues.
side effect of selecting carbon steel is that rust buildup on the
tubes impedes heat transfer and restricts cooling water flow, re- Takeaway. These examples represent a small percentage of
sulting in a reduction in performance. Although carbon steel the potential issues that equipment users encounter when not
works well at first, its performance quickly dissipates, resulting sourcing the correct material or improperly completing main-
in tower instability and poor vacuum. As leaks develop, vast tenance work. Often, the desire to save money and time on the
amounts of cooling water are also lost into the waste stream, front end requires additional work on the back end, which is
increasing the load to the water treatment plant. detrimental to system performance. Turnaround schedules
Due to the constant failure of the carbon-steel bundles, the amplify most vacuum system mistakes, limiting the timely cor-
refinery began replacing them every shutdown going forward. rection of errors and increasing the economic impact of those
Before reaching out to the OEM for assistance, a total of 18 mistakes. While glaring mistakes are often learned the hard
new carbon-steel condenser bundles were purchased from the way, equipment and maintenance costs are often minor when
heat exchanger vendor and installed, far exceeding the cost compared to the cost of lost production and downtime. Care-
of stainless bundles. FIG. 6 shows failed bundles waiting to be fully planning maintenance activities and procuring OEM re-
scrapped. Replacing bundles has additional ancillary costs be- placement parts far in advance of a shutdown helps reduce risk
yond the bundles themselves that include planning, shipping and ensures the continued success of the vacuum system.
costs, cranes, additional manpower and lost production.
While onsite, it was also noted that the replacement bundles ERIC MICHAEL JOHNSON has worked for 7 yr as a Service
were designed incorrectly. The heat exchanger manufacturer Engineer for Graham Corp. in Batavia, New York, specializing
in the evaluation, troubleshooting and commissioning of
was ignorant in their understanding of vacuum condensers. vacuum equipment. He previously acted as the Supervisor of
They utilized multiple features that were incorrect for a vacu- Graham’s service department, overseeing the daily operation
um service. A pressure and temperature survey of the vacuum of its service team and providing support for vacuum systems
globally. He now assists Graham Corp. with its expanding
system highlighted this impact on the tower’s top pressure. The support role by helping to manage the company’s newly opened Gulf Coast Office
system, designed to operate at 10 mmHgA, could only achieve in Covington, Louisiana. Mr. Johnson graduated from Rensselaer Polytechnic
a pressure of 72 mmHgA. The poor tower pressure became sta- Institute in Troy, New York with a BS degree in mechanical engineering.
tus quo, and many of the plant’s engineering staff began to ac-

Electronic and single printed copies for distribution with permission to Graham Corporation from Hydrocarbon Processing
December © 2018 Gulf Publishing Company

Hydrocarbon Processing | DECEMBER 2018