MORE TOWER DAMAGES CAUSED BY WATER-INDUCED WATER-INDUCED PRESSURE SURGE: UNPRECEDENTED SEQUENCES OF EVENTS 1
Ademaro Marchiori , 3
Ana Lidia Wild , 1
Aristides Yoshiaki Saito , 2
Arlan Lucas de Souza , 3
Carmen Mittmann , 3
Christian C. Anton , 3
Felipe Saldanha Duarte , 3
Sandro L.A. Pereira , 2
and Silvia Waintraub 1
PETROBRAS/HEADQUARTERS PETROBRAS/RESEARCH PETROBRAS/RESEARCH CENTER- CENPES 3 PETROBRAS/REFAP – ALBERTO PASQUALINI REFINERY 2
Rio de Janeiro Brazil
Prepared for Presentation at the AIChE 2013 Spring National Meeting, San Antonio, April 28 – May 2, 2013 Copyright © PETROBRAS March/2013 Unpublished
AIChE shall not be responsible for statements or opinions contained in papers or printed in its publications
Abstract Three water-induced pressure surge cases in crude towers are presented. Although different and independent, all the cases have a common point: the water-induced pressure surge is associated to a pressure relief valve opening event. In the first case study, a 126,000 bpd crude distillation unit had a significant decrease in one of the atmospheric tower side withdrawals after an emergency shutdown. The light diesel oil production was cut in less than half. A generalized tower temperature profile reduction was noticed: 27°F (15°C) temperature decrease in light diesel oil pan; 63°F (35°C) in heavy diesel oil pan and 18°F (10°C) in atmospheric residue withdrawal. The second case study occurred at another crude distillation unit located at the same refinery with a capacity of 69,200 bpd. It was observed a deep reduction in the first side withdrawal (naphtha) yield that was compensated by the other side draw products yields. No loss in total diesel production occurred. Compared to the second, the damages in the third case were higher and it was noticed a degradation of products to the bottoms residue. This problem happened at the same tower of the second case, one year later and it was caused by distinct conjugation of facts. The paper presents the troubleshooting techniques applied to evaluate the different units problems (operational tests, gamma scans surveys, and thermograph readings), the root causes of these incidents and the mitigation actions.
Introduction
Crude Distillation Unit (CDU) is present in all refineries. It is the first unit that receives the crude, determines the maximum refinery throughput, minimizes residue production with a lower cost and separates the distillation products as LPG, Straight Run Naphtha, Kerosene and Diesel. A simplified sketch of an Atmospheric Crude Distillation Unit is shown in Figure 1. ATMOSPHERIC TOWER
LPG HEAVY NAPHTHA
KEROSENE
LIGHT NAPHTHA
DIESEL
DESALTER
CRUDE FEED
FIRST TRAIN
SECOND TRAIN ATMOSPHERIC FURNACE
Figure 1
STEAM REDUCED CRUDE
Atmospheric Crude Distillation Unit Sketch
As a CDU operates with very high feed input, a distillation tower malfunction implies in a deep profitability loss. Despite of the huge improvement in distillation technology and knowledge, the number of tower malfunctions increases year after year. Tower internals damage has been the third most common tower malfunction, being most of the time caused by water-induced pressure (Kister, 2003). Although other sources are also observed, the most common water-induced pressure surge is due to water carry over in stripping steam lines. Water accumulation in dead pockets is the fourth cause (Kister,2003). Kister (2006) published a special topic related to water-induced pressure surges, showing cases due to water in feed and slop, accumulated water in transfer line to tower and in heater passes, water accumulation in dead pockets, water pockets in pump or spare pump lines, undrained stripping steam lines, condensed steam or refluxed water reaching hot section and oil entering waterfilled region. Tower damages caused by water-induced pressure surges after pressure safety relief valve (PSV) opening were not easily found in literature and were not common to happen in Petrobras
refineries. A great number of pressure relief valves are installed in a Crude Distillation Unit in order to avoid an overpressure that can occur due to many different causes, such as fire, electric power failure, water to condenser supply failure and valve inadvertent closure, among others (API 521, 2007). There are two major common destinations of the liquid discharge of these pressure relief valves: atmospheric or preflash tower flash zone or a blowdown drum. The CDU units of the majority of Petrobras refineries have the discharge of liquid PSVs to the column flash zone. These PSVs liquid discharges should be composed only by hydrocarbons resulting in no harm to the tower. On the other hand, if water is carried over together with the liquid, a pressure surge can happen inside the column, caused by abrupt expansion of water when exposed to high temperatures. It is mandatory focusing to avoid water accumulation at all steps of the unit cycle life: basic design, detailed engineering, construction, inspection and all the time during operation, especially after shutdowns and start-ups. In the present paper it will be described three case studies, all different and independent, but with a common point: the water-induced pressure surge was associated to a pressure relief valve opening event and caused a severe damage to the tower internals.
First Case Study: PSVs liquid discharge to a pumparound return The simplified flowsheet of the Crude Distillation Unit with a nominal capacity of 126 KBPD is presented in Figure 2. The atmospheric column has four sidedraws (Heavy Naphtha, Kerosene, Light Diesel and Heavy Diesel) and the reduced crude is sent to a RFCC unit. Gas
Crude Steam
First train
Heavy naphtha
Gas
Top pumparound Steam
Desalter Atmospheric Tower
Kerosene LPG
Second train
Steam
Bottom pumparound s stem
Light diesel
Naphtha Steam Heavy diesel Steam
Reduced Crude
Figure 2
Fisrt Case Study – CDU Simplified Flowsheet
In January 2011, an electric power failure resulted in an emergency shutdown. The unit started-up after two days and it was observed that the light diesel oil production was cut in less than half. A
3
generalized tower temperature profile reduction and a 1000 m /d loss in total diesel production was noticed as shown in Table 1. VARIABLE VARIATION LIGHT DIESEL YIELD (m³/d) - 2000 HEAVY DIESEL YIELD (m³/d) + 600 TOTAL DIESEL YIELD (m³/d) -1000 LIGHT DIESEL TEMP. (°C/°F) -15/-27 HEAVY DIESEL TEMP. (°C/°F) -35/-63 REDUCED CRUDE TEMP. (°C/°F) -10/-18 Table 1 Conditions after Emergency Shutdown Thermograph readings and gamma scans surveys were done in order to look for possible damages inside the column.
LIGHT DIESEL NOZZLE
OUTLET LIGHT DIESEL VALVE OPENED
OUTLET LIGHT DIESEL VALVE CLOSED
LIGHT DIESEL OULET LINE
TWO TEMPERATURES LIQUID AND VAPOR
ONE TEMPERATURE LIQUID
Figure 3
Figure 4
First Case Study – Thermograph Readings
First Case Study – Crude Tower Gamma Scan (partial)
As it was possible to draw only half of the light diesel rate, thermografph readings were done close to its draw-off nozzle. The outlet light diesel valve was completely closed and as shown in
Figure 3, the collector channel was able to retain the liquid meaning that probably the light diesel tray itself ( #13) should be harmed. The gamma scan analysis, partially presented in Figure 4, indicated that this tray liquid level was very low, a difference in liquid height between the north and south sides, and also pointed out that the region between the kerosene (# 21) and light diesel draw (#13) had problems. The unit was shutdown in order to repair the internal mechanical tower damages. After inspection it was observed a lot of missing pannels, many opened trapdoors, some trays partially fallen and from observation it was concluded that an uplift from bottom to up had happened. The crude tower sketch is presented in Figure 5. d = 7500 mm
KEROSENE DRAW
.
DAMAGED TRAYS
. PUMPAROUND RETURN
LIGHT DIESEL DRAW
PUMPAROUND OULET HEAVY DIESEL DRAW
FEED
REDUCED CRUDE
Figure 5
First Case Study - Atmospheric Crude Tower Sketch
TRAY # 21
TRAY # 20
TRAY # 13 – LIGHT DIESEL DRAW
Figure 6
First Case Study – Damaged Trays Pictures
Some pictures of the damaged trays are shown in Figure 6. The solid marks represent missing panels and the dashed lines represent an upholstered region. The tower was fixed and after the new start-up the diesel yield returned to the expected values. Investigation to find the causes of the damages showed that a water-induced pressure surge had happened due to water presence during the start-up after the emergency shutdown. Immediately 2 after the heavy diesel pump was aligned it was observed an increase of 0.66 kgf/cm in 30 seconds in top and flash zone tower pressures as it can be seen in Figure 7. It was observed by the operation team that the PSV of heavy diesel X crude exchanger had opened. The liquid discharge of this pressure relief valve is sent to the bottom pumparound system returning to the tower as showed in Figure 8. Unfortunately there are dead pockets that can accumulate water in this circuit and they were not drained because the shutdown was too short. This pumparound return is just below the damaged region of the crude tower and it was noted that the trays major harmed portions were concentrated in the southeast side. Crude Tower Overpressure
Figure 7
Heavy Diesel Pump Start-Up
First Case Study – Overpressure
PSV
Dead Pocket
Top pumparound system
Crude
Naphta or Kerosene Heat Exchanger PSV
Atmospheric Distillation Tower Dead Pocket
Bottom pumparound system Crude
Heavy Diesel or Light Diesel Heat Exchanger
Figure 8
First Case Study – PSVs Discharged to Pumparound Returns
During the start-up operation the distillates pumps work with greater head values, close to their shut off, being common to happen opening of the PSVs that protect these systems. In this procedure the pumparounds pumps begin to operate later on after the tower is already heated. After returning from the emergency shutdown and at the same moment of the heavy diesel pump start-up, the PSV that protects the crude X HD exchanger opened, sending hydrocarbons to the pumparound system returning to the crude tower, carrying water together and causing the overpressure due to abrupt vaporization in high temperatures. Ten months later another water-induced surge damaged the atmospheric tower of another crude unit located at the same refinery and this incident will be the next case study. Second Case Study: PSVs liquid discharge to the atmospheric tower flash zone The simplified flowsheet and control loops of the 69,200 BPD crude distillation unit are presented in Figure 9. There is no preflash drum nor preflash tower. The unit feed rate is controlled by FV-106 located after the booster pumps downstream of the desalters.
HN KER
FV-106 FV-106 Atm crude st
1sr desalter Desalter
Tower
nd 22nd desalter Desalter
2nd train Train
11stst train
Figure 9
Second Case Study – CDU sketch and control loops
In November 2011 a mechanical problem happened to the feed control valve (FV-106) and it was abruptly closed. The charge pump remained sending crude to the unit, the system became over pressurized and the PSVs of both desalters opened, discharging to the crude tower flash zone. The 2 2 crude tower flash zone pressure increased 1.3 kgf/cm and the top pressure 1.1 kgf/cm both in 1.5 min, as shown in Figure 10.
Figure 10 Second Case Study – Crude Tower Overpressure
LD HD
After this event, it was observed a huge decrease in the first side-drawoff production (heavy naphtha), as demonstrated in Figure 11. This reduction was compensated by the lower sidedraws and no diesel loss to the bottoms residue was observed.
Heavy Naphta 3 (m /d)
Figure 11 Second Case Study – Heavy Naphta Yield Reduction A crude tower gamma scan was done in order to show the extent of the damages. It pointed out that only the trays located at the smaller diameter section were with problems. One year later, in November 2012, it was the scheduled unit turnaround and the tower was inspected. Many trays were completely fallen (15 from 44), others partially damaged and from observation it was concluded that an uplift from bottom to up happened. All the damaged trays were sieve type and were located at the tower upper section, in the smaller diameter region. The valve trays located at the larger diameter section were in place. These valve trays opened area are four times the opened area of the upper sieve trays. The crude tower sketch is presented in Figure 12, where it can be seen in blue the valve trays, in red the sieve trays and in highlight the region that was damaged.
UPWARDS DEFORMED TRAYS
#44 #43 #42 #41 #40 #39 #38 #37
TOTALLY OR PARTIALLY FALLEN TRAYS
#36 #35 #34 #33 #32 #31 #30 #29 #28
#27 #26 #25
#24 #23 #22 #21 #20 #19
Sieve Trays Valve Trays
#18 #17 #16 #15 #14 #13 #12 #11 #10
Figure 12 Second Case Study – Crude Tower Sketch
Although the pressure surge occurred at the flash zone, the packing and the valve trays located below the damaged trays and closer to the surge source were preserved. A possible explanation is based in the greater open area and tower diameter in this tower section. Some pictures of the damage trays are showed in Figure 13.
Tray # 29 fallen above tray #28
Tray # 40 fallen above tray #39
Tray # 42 with deformation upwards
Figure 13 Second Case Study – Damaged Trays Pictures
Investigation to find the causes of the incident showed a synchronism between the desalter safety relief valve opening, the atmospheric tower flash zone over pressure and an increase in water presence at the overhead drum. All of the observations led to a conclusion that the tray damages were caused by water-induced pressure surge. The source of this water was attributed to a not drained dead pocket in the second desalter PSV discharge line as can be seen in Figure 14.
Volume
2
nd
Desalter
Atm crude Tower
Figure 14 Second Case Study – Water Accumulation in Dead Pocket
The trays were fixed and the entire distillates yields returned to the normal values. Unfortunately one month after the start-up another water-induced pressure surge happened at the same tower, which will be the third case study of this paper.
Third Case Study: PSVs liquid discharge to the atmospheric tower flash zone
In January 2013, after one month of operation, another incident happened to the same atmospheric tower. Due to a misunderstood interlock action, the feed charge booster pump B-120 A/B showed in Figure 9, was shutdown with continue operation of the feed charge pump B-101 A/B. Again the desalter safety relief valve opened and once more the atmospheric tower flash 2 zone was pressurized. The flash zone pressure increased 0.9 kgf/cm in 20 sec while the top pressure 2 increased 0.85 kgf/cm in 1 min, as can be seen in Figure 15.
Figure 15 Third Case Study – Crude Tower Overpressure
After this event it was not possible to draw neither heavy naphtha nor kerosene. The lower draw-off pans were not able to compensate these losses and the bottom residue increased, meaning reduction of the total distillates yield, as indicated in Figure 16. A general decrease in the tower temperature profile was observed as shown in Figure 17.
Heavy Naphta 3 (m /d)
Figure 16 Third Case Study – Heavy Naphta Yield Reduction
Heavy diesel temp. Light diesel temp.
Pumparound temp. Kerosene temp.
Figure 17 Third Case Study – Temperature Profile
The symptoms after this event were much worse than in case 2 and a new gamma scan (partially reproduced in Figure 18 in blue solid line) also indicated greater tower damage, showing a lot of trays partially damaged and trays from #40 to # 35 completely collapsed. Once again the damage trays were located at the tower smaller diameter section, from tray # 28 to tray # 43.
Figure 18 Third Case Study – Gamma Scan (partial)
Figure 19 shows schematically that many PSVs discharge liquid to the atmospheric tower. The heavy naphtha, kerosene and first pumparound systems have their PSV discharging to the first pumparound return, while for the light diesel, heavy diesel and second pumparound systems the discharges are sent to the second pumparound return, and the desalters (S-102 and S-103) and the preheating trains have their PSVs discharging to the tower flash zone.
st
1 PA
nd
2 PA
HN
Kero
PA
Atm crude Tower
FZ
closed opened
steam injection
LD
HD
PA
st
1
2nd Train
Figure 19 Third Case Study – Safety and Relief Valves (PSV) discharges A multi-disciplinary team was designated to find the reasons of the tower damages. Once more the conclusion was water-induced pressure surge, but this time, the water source was found and was still in action during a visit to the operational site just two days after the incident. During the site visit some bad practices were identified to be occurring simultaneously, which can be understood with Figures 19 and Figure 20:
V-117 is a small drum used to indicate when any of these safety relief valves opens. Unexpectedly the level instrument connections were blocked and in fact there was no liquid indication; During the incident investigation it was drained a lot of clean water from V-117 for more than 30 min; This water source was caused by a non blocked steam injection to a valve used only with the purpose of PSV liberation . This valve is highlighted in dashed red line in Figure 20. Steam was flowing since the start-up and continued even after the tower damage event. Although this PSV has a double valve block, at that moment, only one valve was closed but allowing steam passage, and the second one was open due to a difficult access; The spectacle blind of this PSV steam injection system was not closed as it should be. As a matter of fact it can be seen from Figure 19 that all spectacle blinds located at the systems that discharge to the tower flash zone were open.
st 1 Desalter
2nd Desalter Atm crude Tower
steam injection
Volume
Figure 20 Third Case Study – V-117 LAH not aligned and steam injection not blocked
Conclusions Towers malfunctions and as consequence losses in distillates yields and profitability, although undesirable, unfortunately are very common. Three case studies of tower internals damage caused by water-induced pressure surge were presented. All the cases have in common an atmospheric crude tower overpressure after a pressure relief valve opening event, discharging hydrocarbons and carrying water into the tower. Although all the cases are related to PSV opening and liquid discharging to the tower, the problem exists only due to the water presence. It should be mentioned that in this CDU many other times the desalter PSV opened discharging to the tower flash zone (in fact 16 times), but with no harm. In all of these events no increase in water accumulation was observed at the overhead drum. It is mandatory to avoid water presence at the PSVs collectors that discharge liquid to towers. Focusing in eliminating water accumulation is necessary in all steps of the unit cycle life: basic design, detailed engineering, construction, inspection and all the time during operation, especially after shutdowns and start-ups. The liquid collector should be inclined with no dead pockets. There must be a reliable way of identifying liquid presence. Drainage is very important. A more conservative design, but more expensive, is to have a blowdown drum collecting all the PSVs liquid discharges.
The case studies are summarized in Table 2.
Initial event PSV location PSV discharge
Case Study 1 heavy diesel pump start-up after emergency shutdown crude X diesel heat exchanger pumparound atmospheric tower return line
Case Study 2 feed charge valve closure desalter atmospheric tower flash zone
Water source
possible undrained dead pocket
possible undrained dead pocket
Mitigation
avoid dead pockets during design and operational drainage routine
avoid dead pockets during design and operational drainage routine
Case Study 3 feed charge booster pump shutdown desalter atmospheric tower flash zone unlocked PSV liberation steam valve and not aligned level indicator operational training
Table 2 – Case Studies Summary
References American Petroleum Institute, 2007, Pressure-relieving and Depressing Systems, ANSI/API Standard th 521 5 Edition, Addendum May 2008. Kister, H.Z., 2006, Distillation Troubleshooting (John Wiley & Sons, Inc., New Jersey, USA). Kister, H.Z. 2003, What Caused Tower Malfunctions in the last 50 years ? Trans IChemE, Vol 81, Part A, January 2003.