Steam-surface condensers: Lessons from the field

J. Lines, Graham Corp., Batavia, New York

Steam-surface condensers are used throughout the refining, petrochemical and chemical industries (FIG. 1). The two principal applications are:

1. Mechanical drive turbine exhaust, where the turbine is powering a compressor, blower or pump
2. Turbine-generator set for electrical power production.

In either service, the steam-surface condenser maintains a sub-atmospheric or vacuum condition at the turbine discharge. This enables the turbine to effectively create more power than if the turbine exhausted to atmospheric pressure.

A condenser is seemingly straightforward static equipment: in one respect, this is true, as it condenses steam with water typically as the coolant; conversely, a condenser is rather complex, as it is the heat sink for a turbine that is always considered critical equipment inside battery limits or within the utility plant. If a turbine underperforms due to poor surface condenser performance, plant throughput is affected negatively—therefore, the steam-surface condenser will often be considered critical equipment for ethylene, methanol, ammonia, nitrogen, propane dehydrogenation and other process services.

This complexity arises from the steam-surface condenser along with its ejector venting package is what maintains the turbine power output capability by achieving design backpressure on the turbine. The operating pressure of a steam-surface condenser is under vacuum, and there is air in-leakage into the system—thus, a low pressure-drop condensation under vacuum conditions (with the necessary gathering, subcooling and continual evacuation of non-condensable gases from the condenser bundle) makes this design and construction complex and quite different from conventional shell-and-tube heat exchangers.

This article details the complexity of this static equipment, along with the consequences for the end user when the performance of a steam-surface condenser is not at design.

Steam-surface condenser. This type of condenser receives steam exhaust from a turbine. The steam is condensed on the outside surface of the tubing, with water used as the coolant within the tubes. Condensation is essentially isothermal, owing to only a small quantity of non-condensable gases within the turbine exhaust steam. The Heat Exchange Institute (HEI) provides critical design information for this type of condenser.

The condenser is what creates the vacuum condition at the discharge of the turbine, while the ejector venting system that supports the condenser is designed to continuously evacuate non-condensable gases from the condenser. Under normal circumstances, and when performance is as expected, the ejector system will not set the operating pressure of the condenser. An ejector system can, and will, set the operating pressure if something is wrong, such as high air leakage, poor motive steam conditions or a mechanical issue with the ejector system (FIG. 2).

The condenser is a critical heat sink for the turbine exhaust steam. Provided the ejector system is not establishing the operating pressure of the condenser, the condenser pressure is established via thermal equilibrium and from balancing the thermal duty equation: Q = U*area (A)*logarithmic mean temperature difference (LMTD). Once a condenser is installed, the heat transfer surface area is fixed. The thermal duty, total heat transfer rate and LMTD can vary. TABLE 1 is directional in nature, describing how condenser pressure will vary as operating variables change.

The author’s company routinely surveys condenser installations to provide operating assessments, troubleshoot underperformance, train end user operators, and provide recommendations regarding performance risk mitigation and opportunities to improve performance and operational reliability, or to lower energy consumption.

The following are five case studies to identify the root causes for condenser underperformance and the resultant plant throughput constraints for the end user.

Caribbean ammonia plant. An ammonia plant in the Caribbean had been in operation for approximately 40 yr. The turbine driving the syngas compressor was operating at an elevated backpressure due to an undesirable increase in the operating pressure of the steam-surface condenser that condenses the turbine’s exhaust flow. The condenser was the original supply, and the last plant turnaround for maintenance had been 10 yr prior to the 2019 visit by the author’s company.

This original surface condenser was supplied in 1980 and was designed to maintain a backpressure of 11.86 kPa (1.72 psia) at the steam turbine exhaust. The unit was cooled by seawater and was designed for inlet and outlet temperatures of 28.9°C (84°F) and 46.1°C (115°F), respectively. The designed heat load to the condenser was 293 MMBtu/hr (85.9 MW).

Plant data showed that the pressure maintained by the condenser was 38.6 kPa (5.6 psia), which was well above the design basis (FIG. 3). Furthermore, the turbine exhaust flow was about 50% above design, as the plant pushed more steam through the turbine to compensate for lost generated power, which was attributed to the elevated turbine exhaust pressure. The cooling water supply temperature to the condenser was 30.6°C (87°F), which also negatively impacted the operating pressure—it was 1.7°C (3°F) warmer than the design basis. Seawater flow was comparable to the design basis.

The excessive turbine exhaust flow was 50% higher than the design basis, which was a cause for the elevated operating pressure. Upon deeper analysis, the thermal duty was up approximately 50%, due to a 50% higher turbine exhaust flow, while the LMTD was greater by more than a factor of three times. The designed LMTD was 16.4°F, while the actual observed LMTD was 30.6°C (55°F). This suggests that the overall heat transfer coefficient in operation was well below the design basis. This implies a potential fouling issue. The condenser had 3,716 m² (40,000 ft²) of surface area. The Udesign was 2,175 kcal/hr m² °C (446 Btu/hr ft² °F). Uworking was approximately 975 kcal/hr m² °C (200 Btu/hr ft² °F).

The installation was shut down, and the condenser internals were opened for inspection. It was identified that tube-side fouling and tube blockage had occurred, thereby reducing the overall heat transfer rate. It had been more than 10 yr since the condenser was cleaned. FIG. 4 shows the solidified deposits on the tubes, along with blocked tube holes. In addition, the condenser water box (channel) had corroded due to a loss of the coating, which led to seawater corrosion (FIG. 5). This corrosion resulted in a loss of seal between the first and second water passes in the condenser, leading to bypassing and maldistribution.

The author’s company recommended the following:

1. Considering that the unit had operated for 40 yr and was reaching its end of life, it was recommended that the condenser be replaced. It was too risky to provide a bundle kit consisting of new tubesheets, tubing, support plates and tie rods/nuts.
2. At a minimum, the water boxes should be replaced, and the tube side of the condenser should be cleaned.
3. A new baseline performance assessment should be established.
4. A routine performance assessment should be conducted and compared against the baseline.
5. Maintenance and internal inspections should be performed at each shutdown, and the operator was advised not to attempt to run for > 10 yr without the necessary turnarounds for performance maintenance.

U.S. Gulf Coast ethylene plant. During a 2016 outage, an ethylene producer on the U.S. Gulf Coast replaced a cracked-gas compressor turbine exhaust condenser that had been in operation since 1990. Upon bringing the plant back online, it was not possible to achieve production capacity due to the cracked-gas compressor’s performance. The horsepower generated by the turbine was below design due to the replaced steam-surface condenser operating at 19 kPa (2.75 psia) rather than the designed 15.2 kPa (2.2 psia).

Because the condenser was newly installed, fouling on the tube side was not suspected. The plant requested a site visit by the author’s company. The engineer assessed the installation and completed a set of field measurements to compare to the design basis (FIG. 6).

The field measurements confirmed an elevated operating pressure by the steam-surface condenser. Several options were available to determine if the condenser was setting the elevated pressure or if it was being caused by the downstream ejector system used for continuously removing non-condensable gases from the condenser. The engineer turned on the hogging ejector—which is used only at startup to quickly evacuate air and establish a vacuum condition—while running the evacuation ejectors. When this was done, the condenser pressure was reduced to 12.9 kPa (1.8 psia). This quick test confirmed that the condenser was not thermally limiting performance. It was the ejector system venting package that was causing the elevated operating pressure.

If the ejector system is a cause for poor performance (such as elevated pressure), various potential causes include:

1. The ejector motive steam conditions are below the design basis (which was not the case here).
2. There are mechanical issues with the ejector system due to corrosion, erosion, welding or casting cracks, or misalignment.
3. High levels of air leakage.

To assess air leakage, an air leakage meter on the after-condenser vent can be used; however, one was not available in this case. While not sophisticated, a 33-gal industrial trash bag was used. It was placed on the after-condenser vent, and the time to fill the bag was recorded. This simple test confirmed that air in-leakage was 2.25x the design level. This excess air leakage overloaded the venting ejector system, causing pressure in the condenser to rise. The ejector system was setting the pressure—an unwanted condition under all circumstances.

It can seem odd that a rather small ejector system downstream of a large condenser can cause elevated pressures to occur. This can be called “the tail wagging the dog.” The little ejector system (vs. the large surface condenser) can, and will, establish the operating pressure for the condenser when that condenser can no longer thermally establish the operating pressure.

Having identified excessive air leakage as the root cause, it became necessary to find the location of the leakage. Third-party helium leak testing was performed. It was identified that the expansion joint between the turbine exhaust and surface condenser had cracked at the bellows, permitting air to leak into the system, as it was operating at sub-atmospheric conditions.

Upon repairing the cracked bellows, air leakage was restored to normal, the condenser operating pressure held at satisfactory pressure and the turbine-compressor performance enabled the ethylene plant to run at capacity.

Latin American nitrogen plant. A nitrogen production plant in Latin America had replaced an original surface condenser—which started operations in 1966—with a design by a shell-and-tube company rather than by the original equipment manufacturer. This replacement was completed in 2011. While steam-surface condensers fall under the broad umbrella of shell-and-tube heat exchangers, they are designed very differently due to sub-atmospheric operating conditions and internal configurations to collect, subcool and evacuate non-condensable gases that inevitably enter the process due to the condenser’s operation under vacuum.

After trying to understand and resolve the performance issues with the shell-and-tube supplier, the end user reached out to the author’s company. The designed operating pressure of the turbine exhaust condenser was 12.7 kPa (1.85 psia), but the new condenser was achieving 33.9 kPa (4.9 psia). At such an elevated pressure, the steam turbine was able to run at only 60% of its designed power output capacity—thus, the plant’s production capacity was seriously below nameplate capacity. The consequence of the elevated pressure was less power generated from the turbine that drives the critical process compressor, resulting in lower plant production capacity. To rectify this situation, the end user needed assistance quickly. An engineer was dispatched to the site to collect performance data, review drawings and develop a course of action (FIG. 7).

A complete performance survey was not undertaken for the ejector system venting package. A no-load test was completed by isolating the ejector system from the surface condenser to evaluate pressures when exposed to no air load. The no-load test confirmed that the ejector system was performing properly.

The engineer requested a review of the condenser drawings, which showed that the nameplate on the condenser was that of a shell-and-tube supplier and not a company considered a steam-surface supplier.

It was identified that the shell-side tube bundle design was the cause for the elevated operating pressure. Two design concerns were identified. First, there was a perforated plate running the full width and length of the tube bundle atop the tube field. The engineer envisioned that this was done to ensure uniform steam flow distribution into the tube field of the condenser. However, this resulted in a high pressure drop as the steam passed through the perforations, causing the condenser’s operating pressure to rise in response. Secondly, there was no location within the tube bundle for separating the non-condensable gases from the condensate, gathering and subcooling the gases, and ultimately extracting them from the condenser. The shell-and-tube company did not appropriately consider the venting of non-condensable gases within the condenser. This elevated the operating pressure, as well.

FIG. 8 is a depiction of the shell-and-tube supplier’s design using a perforated plate for flow distribution, which resulted in a large pressure drop and an elevated operating pressure. This approach by the shell-and-tube supplier is very different from the methods used by surface condenser suppliers to create uniform bundle penetration and to gather non-condensable gases separated from the condensate for subcooling and extraction/evacuation from the condenser, as shown in the steam-surface condenser model in FIG. 9.

The engineer’s recommendations were to replace the recently installed condensers, since little could be done to alleviate the performance shortfall due to the perforated flow distribution plate and a lack of non-condensable-gas handling within the condenser. The shell-and-tube company did not design the condenser for operating under vacuum or for venting the condenser.

Middle East aromatic plant process. A benzene, toluene and xylene (BTX) aromatics plant in the Middle East had a surface condenser serving a turbine powering a recycle gas compressor. The condenser started operations in the late 1990s. Between the first and second turnarounds, it was noted that many tubes in the surface condenser failed, developing leaks and those tubes had to be plugged. The tubing experienced corrosion, which led to cooling water leaking into the steam side of the condenser.

This condenser had been in operation for about 15 yr. The end user requested that the author’s company visit the site to evaluate the situation and aid the third-party contractor that was to remove the failed tubes from the condenser.

Upon arriving at the site and inspecting the tube sheet of the condenser, it was evident that the tubes in the air-cooling section of the condenser had failed and were plugged. The tubing material was admiralty brass, and the tubes only failed in the air-cooling section (FIG. 10).

This type of failure can occur with brass tubing when ammonia is present. Boiler treatment chemicals can introduce ammonia into the steam cycle. Normally, this does not present issues; however, in the air-cooling section of a steam-surface condenser where the concentration or mole fraction (partial pressure) of ammonia and oxygen is highest, these gases can dissolve into the condensate, causing it to become corrosive. Where the brass tubing passes through the support plates, the corrosive condensate is continually wetting this area, causing a condition called “condensate grooving” on each side of the support plate.

The engineer recommended replacing the admiralty brass tubes in the air-cooling section with material better suited to resist this form of corrosion. The recommendation was to install 316SS or alloy 2205 duplex tubes in the air-cooling section. The thermal conductivity of the suggested material is inferior to that of admiralty brass; however, the air-cooling section is generally less than 10% of the overall tubes in a surface condenser. Therefore, the greater thermal resistance of stainless steel or duplex stainless steel will not materially impact condenser performance, but will eliminate the corrosion experienced by the brass tubes.

U.S. purified terephthalic acid (PTA) plant. A U.S. PTA plant had begun commercial operations and had a 30-MW turbine-generator producing electrical power from internally sourced waste steam. Since startup, an ongoing issue was unstable turbine exhaust pressure due to steam-surface condenser operating pressure rhythmically varying between 6.9 kPa and 4.8 kPa (1 psia and 0.7 psia). The operating condition of the turbine was unstable, and the variation in turbine exhaust pressure was unacceptable.

The site shared a trend of condenser pressure vs. time that highlighted this unusual and rhythmic pattern (FIG. 11). The author’s company was dispatched to the site to assess the installation and to gather operating data. An issue with the steam ejector evacuation system was suspected. FIG. 12 details the site’s design basis vs. field measurements.

It was observed that the pressure between the first- and second-stage ejectors was very low, which indicated that the overall system was very tight with no air leakage. The pressure between the ejectors was anticipated to be in the range of 2.8 psia with normal levels of air leakage into the system. The measured data was 1.3 psia.

The engineer then took the system offline to perform no-load testing on the ejectors. It was identified that the second-stage ejector was unstable at no-load. This means that it was unable to maintain its compression shockwave, and it cycled between operating correctly and losing shockwave, which would lead to a rise in surface condenser pressure.

Last-stage ejectors in a surface condenser venting package can be designed for no-load stability based on the diffuser style and the amount of motive steam. It is rare to not have air leakage into a process under vacuum, but, in this instance, there was none. In this case, the ejector was already installed, and the time to provide a replacement ejector was unsatisfactory for the customer. While typically counterintuitive, the solution for providing stable operating pressure was to bleed a small amount of air into the suction side of the steam ejector venting package. A valve was added to permit ambient air to be drawn into the system so that the second-stage ejector would operate in a stable condition. Only a very small amount of air was needed; however, upon adding a little air load into the system, the turbine exhaust pressure became stable.

Takeaways. While surface condensers are often considered to be an accessory to steam turbines, they are critical for achieving satisfactory turbine performance. This seemingly straightforward heat exchanger is quite complex to design and construct, and to achieve desired performance. End users can learn numerous lessons from performance improvement engineers in the field. Four of the examples provided in this article are rather common when end users are experiencing elevated operating pressures. The fifth example—where the condenser operating pressure rhythmically varied between 6.9 kPa and 4.8 kPa (1 psia and 0.7 psia)—was unusual.

It is recommended to conduct a baseline operating survey of the surface-condenser system shortly after installation during warm summer months when the system should be experiencing its most challenging operating conditions. Operators should also periodically assess performance against benchmark data. If there is a noticeable and undesirable change to operating pressure, a surface condenser performance improvement engineer should be brought in to aid the system in getting back to designed performance. Several variables are at play when it comes to condenser performance, and operators can quickly troubleshoot challenges by bringing in experts.

The planning of a scheduled turnaround should permit a performance survey approximately 15 mos–18 mos before the shutdown. This will permit time to analyze current performance, to vet opportunities for improving performance or for returning to designed performance, and to fabricate replacement equipment on a non-urgent timeline. HP

JIM LINES is an engineer recently retired from Graham Corp. He worked at the organization for 37 yr, and has authored numerous articles related to heat transfer, vacuum process condensers and ejector systems. He has also held various engineering and management positions at Graham Corp.