naRTC 2026
Demystifying vacuum ejector systems
Scott Golden, Tony Barletta, and Steve White Process Consulting Services, Inc.
Intercondenser thermal behaviour The first-stage intercondenser removes most of the steam and condensable oil load. Only a small fraction of vapour exits to the sec- ond stage, and a disproportionate amount of condensation occurs in the sub-cooling zone. Although this zone accounts for only about 5% of the exchanger duty, it can condense several thousand pounds per hour of steam. Loss of this duty due to bypass dramatically increases the second-stage load. Vacuum systems are highly sensitive to CW temperature and flow. Small increases in CW temperature or fouling can raise operating pressure, but if the long air baf- fle is intact, sub-cooling is maintained and vapour temperature remains below conden- sate temperature. This distinction is critical as general fouling and long air baffle bypass have different thermal signatures. Cracked gas rate and design margin Designing vacuum systems with unrealis- tically low cracked gas rates and minimal MDP margin reduces steam consumption but leaves no tolerance for fouling, CW variation, or mechanical defects. Systems designed with higher cracked gas allowances and ade- quate MDP margin can tolerate some loss of sub-cooling without immediate performance break. Lower actual cracked gas rates reduce vapour carryover, effectively increasing mar- gin and allowing higher outlet temperatures without overloading downstream ejectors. Second-stage ejector sensitivity Second-stage ejectors typically have lim- ited capacity to handle the incremental load associated with a poorly performing first- stage intercondenser. The incremental load can consume all available margin and cause a first-stage performance break, illustrating how even a minor long air baffle bypass can have system-wide consequences. Conclusion Vacuum ejector systems are commonly designed with minimal margins to reduce capital cost and utility consumption. Modern X-shell intercondensers with sub-cooling zones provide significant benefits, but only if they function exactly as intended. Failure of the long air baffle to seal properly allows vapour bypass, overloads the second-stage ejector, and frequently results in a first- stage ejector performance break. Vapour outlet temperature higher than condensate temperature indicates sub- cooling failure. Unfortunately, inadequate instrumentation often leads to incorrect diagnoses. To reduce risks of a performance break, ejectors should be shop tested to generate certified performance curves and verify true MDP. Adequate design margins should be applied to cracked gas rates, con- denser fouling, CW conditions, and MDP. While some operational variability is una- voidable, a performance break is often a design and mechanical issue. A robust inter- condenser design, proper execution, and realistic design margins are essential for reliable vacuum system performance.
Vacuum systems are critical to maintain- ing low column operating pressure, maxim- ising distillate recovery, and preserving the economic value of refinery vacuum units. Despite this importance, vacuum ejector systems frequently perform poorly, even immediately after installation or following turnarounds. A common failure mode is a performance break, which can raise col- umn pressure by 20-40 mmHg and sharply reduce distillate yield. Performance breaks occur when the first-stage ejector operates at or above its maximum discharge pressure (MDP). Although there are many causes of poor vacuum system performance, a fre- quent and often overlooked cause is failure of the first-stage intercondenser to sub- cool the outlet vapour ( Figure 1 ). This article focuses on the importance of the first-stage intercondenser, particularly its sub-cooling zone. It explains how failure of this function commonly leads to perfor- mance breaks and unstable, elevated col- umn pressure. Fundamental concepts such as ejector performance curves, MDP, con- denser design, and component interaction are reviewed, along with critical design- phase considerations, including cracked gas rates, MDP margin, and system pres- sure drop. Ejector system design and interaction Most refinery vacuum systems use three- or four-stage ejector trains. Each stage con- sists of a steam ejector followed by a con- denser. Motive steam supplies the energy to compress vapour from the vacuum col- umn top pressure to the outlet pressure of the ejector. Each condenser removes con- densed steam and hydrocarbons, reducing the load to downstream ejectors. Cooling water (CW) rates are primarily set by the first-stage intercondenser, the largest and most critical exchanger in the system. Design optimisation typically focuses on minimising motive steam and CW consump- tion. The first-stage ejector can consume up to 70% of the total motive steam and often represents more than half of the installed cost of the vacuum system. Consequently, first-stage design conditions dominate sys- tem optimisation and represent the greatest opportunity for savings, but also the highest risk if margins are inadequate. Ejector performance is defined by com- pression, entrainment, and expansion ratios, which relate suction pressure, discharge pressure, and motive steam conditions. For a given suction load and pressure, the required motive steam increases with com- pression ratio (discharge pressure divided by suction pressure). Higher MDP designs consume more steam, so suppliers often minimise MDP to reduce utility consump- tion. Since end users frequently provide little guidance on required design mar- gin, systems are commonly designed with extremely thin margins. As a result, even small deviations from design conditions or underperformance of the first-stage inter- condenser can trigger a performance break.
design data rather than actual field meas- urements can lead to incorrect conclusions. Effective troubleshooting requires pre- cise measurement of small pressure drops (often only 3-6 mmHg across the first-stage intercondenser) and accurate tempera- ture readings. Even small thermocouple errors can obscure the temperature differ- ences that indicate sub-cooling failure. As with pumps, ejector performance cannot be evaluated without knowing actual flow rates, pressures, and performance curves. First-stage ejector suction load definition Proper vacuum system design begins with defining first-stage ejector suction loads. The vapour leaving the vacuum column con- sists of process steam, condensable hydro- carbons, and cracked gas/air leakage. These loads are converted to steam or air equiva- lents using HEI or DIN methods so they can be compared with ejector test curves. Process steam typically dominates the load and includes stripping steam, heater velocity steam, and water carried with the feed. It is usually well defined during pro- cess design. Condensable hydrocarbon load depends on column top temperature and pressure, steam usage, and stripping efficiency in the atmospheric crude col- umn. Poor stripping efficiency can signifi- cantly increase condensable load and lead to undersized vacuum systems. Cracked gas generated in vacuum heater tubes is much harder to estimate accu- rately. It depends on heater design, firing severity, residence time, crude type, and outlet temperature. Although cracked gas typically represents a small fraction of total load, it strongly affects condenser perfor- mance and should be assigned a generous design margin. Ejector performance and MDP Ejector performance curves define the rela- tionship between process load, suction pressure, and MDP. As long as discharge pressure remains below MDP, suction pres- sure is governed by process load. However, once discharge pressure exceeds MDP, suction pressure rises sharply and becomes unstable, marking a performance break. Certified performance curves based on shop testing are essential. When full-scale testing is impractical, scaled replica testing should be required. Field testing has shown that actual MDP can be 15 mmHg lower than vendor-stated values. In some cases, refiners have had to modify ejectors, such as moving motive steam nozzles, to recover lost MDP, frequently at the expense of suc- tion capacity.
1st stage intercondenser long air ba e bypass
1st stage intercondenser vapour outlet temperature is high
2nd stage ejector suction load increases
2nd stage ejector suction pressure increases
1st stage ejector discharge pressure exceeds MDP
1st stage ejector performance break
High vacuum column operating pressure
Figure 1 Cause and effect first-stage ejector break
typically X-shell designs incorporating a dedicated sub-cooling zone created by a long air baffle ( Figure 2 ). This configura- tion evolved from older E-shell designs and offers lower pressure drop and reduced steam consumption. Roughly 20-25% of the tube bundle is dedicated to sub-cooling. The long air baffle forces outlet vapour to pass over tubes cooled by the coldest CW, sub-cooling the vapour below the bulk con- densate temperature. This sub-cooling zone is essential because it minimises the vapour load to the second-stage ejector. The MDP of the first- stage ejector is selected assuming that the intercondenser performs as designed. If sub-cooling fails, additional steam exits the condenser as vapour, increasing the load on the second-stage ejector. This raises sec- ond-stage suction pressure, which, in turn, raises first-stage ejector discharge pres- sure. Once discharge pressure exceeds MDP, a performance break occurs. The most common root cause of sub-cool- ing failure is mechanical bypass of the long air baffle due to poor sealing, corrosion, damage, or installation defects. When the MDP margin is small, as is typical, a perfor- mance break becomes inevitable if bypass occurs. Thus, proper mechanical design, installation, and maintenance of the long air baffle are critical. Diagnosis and troubleshooting challenges The long air baffle bypass is relatively easy to diagnose conceptually: the vapour leav- ing the first-stage intercondenser should be cooler than the condensate leaving the shell. If the vapour temperature is higher, bypass is occurring. In practice, however, troubleshooting vacuum systems is difficult because instrumentation can be inadequate. Flow meters, reliable temperature meas- urements, and accurate pressure data are often missing. Analyses based on original
1st stage ejector vapour inlet
Cooling water outlet (typically 10˚F higher than inlet)
Tube bundles
Vapour outlet
Tube support
‘Long air ba e’
Cooling water inlet (coldest)
Condensate outlet
Importance of first-stage intercondenser Modern first-stage intercondensers are
Contact: TBarletta@revamps.com
Figure 2 X-shell long air baffle
9
Powered by FlippingBook