Roy T. Halle
Exxon Chemical Co.
Baytown, Tex.
Mo Vadekar
Exxon Chemical
Holland By Rotterdam
During early operation of one of Exxon Chemical Co.'s ethane cracking plants, a temperature runaway in a small shell-and-tube heat exchanger upstream of the hydrogen methanator reactor resulted in rupture of the exchanger shell.
Exxon has concluded that the overtemperature resulted from the exothermic heat of reaction of ethylene and hydrogen. This hydrogenation reaction unexpectedly initiated at a temperature well under 300 C.
BACKGROUND
The temperature runaway occurred when level instrumentation on the upstream separator drum malfunctioned, allowing overflow of ethylene-rich liquid into the hydrogen (H2) vapor stream.
This should not have occurred in the absence of a catalyst. Exxon formed the hypothesis that a catalyst was present, to explain the unexpected runaway. It was further hypothesized that the catalyst was finely divided iron or iron oxide from rust, which are known hydrogenation catalysts.
This rust is believed to have been present in the system prior to start-up. Such rust is known to exist as very finely divided particles. Based on experience with iron oxide catalysts, it is likely that the rust was activated by reduction with the hydrogen methanator feed prior to the ethylene carryover.
In addition, there was evidence of unexpected exothermic hydrogenation of ethylene in the main process gas dryers during regeneration whenever the ethylene content of the mostly hydrogen regeneration gas rose above normal. Exxon believes this hydrogenation was also catalyzed by rust or iron from rust.
Laboratory experiments were carried out, the results of which support these beliefs.
This article will cover these relevant features of the process: the methanator preheat exchanger temperature runaway incident, the unexpected temperature increases around the drier circuit, the background which led to the iron rust catalyst hypothesis, and the laboratory work which supports the hypothesis.
The extensive countermeasures taken to permit safe operation of the plant following this incident are not fully covered here.
PROCESS DESCRIPTION
Cracked gas from ethane pyrolysis contains mainly hydrogen, ethylene, and unreacted ethane, plus some methane and small quantities of acetylene, oxides of carbon, and other minor components. In this plant, the product ethylene is purified by cryogenic fractionation in a demethanizer first process sequence (Fig. 1), with "back end" acetylene hydrogenation on the feed to the ethylene splitter.
Hydrogen is separated from the cracked gas in the demethanizer feed chilling train. A small hydrogen slip stream is further purified to about 98 mole % for use in the acetylene hydrogenation step (Fig. 2). This slipstream is purified by further chilling and the addition of ethylene as an absorber oil for methane.
This step is followed by methanation of the minor amount of carbon monoxide (CO) contained in the purified hydrogen stream (Fig. 3). At the very low temperatures in this cryogenic purification step, there is the risk that even traces of acetylene present in the streams will freeze out and accumulate.
The accumulation of acetylene could result in plugging of core exchangers. In addition, there is the concern that spontaneous decomposition of accumulated acetylene will cause equipment damage.
To assure that no acetylene can accumulate in the coldest portion of the plant, ethylene was selected as the absorber oil to be injected in the feed to the purified hydrogen separator drum. Ethylene was selected over ethane because, compared to ethane, it is a much better solvent for acetylene at the process conditions.
The design fully anticipated that if ethylene carried over in the hydrogen from the separator drum, the methanator catalyst would catalyze the hydrogenation of ethylene-a highly exothermic reaction (-33 kcal/g). Carryover and hydrogenation of just 1 mole % ethylene in the 98% hydrogen stream results in a temperature rise of about 50 C.
The hydrogen separator drum liquid is over 70 mole % ethylene. The liquid ethylene pumped into the drum to prevent acetylene precipitation represents the major source of liquid entering the drum, and essentially all the ethylene entering the drum.
The design provided substantial safeguards against the carryover of ethylene to the methanator. The separator drum was provided with three separate level devices, The first was for level control, the second actuated a highlevel alarm, and the third shut off the flow of liquid ethylene to the drum.
The methanator reactor was equipped with bed temperature recorders, high-temperature alarms, and a hightemperature trip system. The trip system consists of six thermocouples distributed through the reactor catalyst bed, and one in the effluent line.
Any one of these seven temperature measurements initiates a reactor trip upon detecting high temperature.
The system was also provided with a manual push button in the control room to initiate the trip. Since the design did not anticipate an exothermic reaction upstream of the catalyst bed, the trip was not initiated by high temperatures in the upstream heat exchange equipment.
On initiation, the trip system automatically operates valves which isolate and bypass the reactor system and depressure the reactor system to the flare. Thus, within a very short time of trip initiation, the reactor would be isolated from the source of reactants, arresting any temperature rise. It would also be fully depressured to flare line pressure, relieving the pressure stress on a possibly overheated reactor.
Reactor feed was heated first in a small shell-and-tube type feed/effluent exchanger, followed by a double pipe 600-psig steam heater. The unexpected runaway temperature occurred in the reactor feed/effluent exchanger on the shell side, which handled reactor feed. The shell failure was nearer the feed inlet than the outlet.
TEMPERATURE RUNAWAY
The incident occurred shortly after initial plant startup. The plant had been in production for about 1 week, and the operation was still being debugged.
With no warning, the exchanger shell ruptured, releasing gas at about 450 psig. The gas ignited. Two technicians in the field immediately reported a fire in the methanator area to the control room by radio.
The control room technician initiated the methanator trip system with the manual button, isolating and depressuring the methanator, including the ruptured exchanger shell. Flames, initially estimated at about 30 ft in height, died down to about 3 ft within 5 min of the rupture, and to about 1 ft within the next 5 min, as the system depressured.
The exchanger was then isolated with manual block valves and the fire went out. No one was injured and damage was limited to the small ruptured exchanger and a minor overhead cable tray. It was subsequently determined that the plant was operable without the damaged exchanger, and the cable tray was easily repaired.
INCIDENT INVESTIGATION
A full investigation was initiated. Efforts concentrated on determining the causes of the problem and designing and implementing corrections to assure future safe operation before the plant was restarted.
Based on analysis of the extensive data available following the incident, it was concluded that the hydrogen separator had filled with liquid and overflowed, although none of the three separator level instruments had detected a high level.
Records show that the liquid outlet level control valve began closing over 40 min prior to the exchanger failure, and was fully closed 35 min prior to the failure. Plant debugging efforts had included work on problems with these level instruments up to the time of the incident.
Following the incident, it was determined that these level instruments had been unreliable because of liquid accumulation in the impulse lines. All three level transmitters shared common impulse lines. Thus, all three were reading the same incorrect level at the time of the incident.
Design modifications and procedures were implemented to ensure that this liquid level control problem is not repeated. Reactor system automatic temperature-runaway protection was extended to include the reactor preheat system.
The postincident investigation showed that about 35 min before the rupture, the temperature downstream of the feed/effluent exchanger started to climb. It reached the top of its range (400 C.) 15-20 min before the rupture and was climbing at a rate of roughly 15 C./min.
It should be noted that this was not the bed inlet temperature control point or a point that the operator would normally monitor. It was provided for exchanger performance monitoring.
At about this time, the ethylene-in-hydrogen analyzer, which had been reading at its base line (less than 0.1%), showed a reading of over 2%. It climbed to over 6%, then dropped to the base line. This drop apparently occurred because the ethylene composition was so far above the normal range of the analyzer.
Subsequent metallurgical examination confirmed overheating as the cause of the exchanger failure. This, together with other plant instrument data, demonstrated conclusively that the exchanger rupture was caused by excessive shell temperature.
Hydrogenation of ethylene is the only reasonable explanation for the abnormally high temperatures.
The methanator inlet temperature rose from about 260 C. before the incident, to only about 290 C. at the time the exchanger ruptured. This was a result of three factors: the transfer of heat to the reactor effluent in the feed/effluent exchanger, the heat capacity of the equipment, and heat losses.
Methanator top-bed temperatures never budged. Thus, no methanator alarms or trips were activated. The lack of temperature rise in the catalyst bed also implies that any ethylene present in the stream had already been reacted before entering the bed.
The next mystery of the incident was this: Why did the hydrogen and ethylene react upstream of the methanator reactor?
Process temperatures were probably too low for the hydrogenation reaction to proceed in the absence of a catalyst.
At no time during the course of the incident did the indicated methanator outlet temperature (the hot fluid on the tube side of the failed feed/effluent exchanger) exceed 250 C. Prior to the incident, the exchanger was heating feed from about 5 C., and the operation was steady. The exchanger shell failed at about one third of the distance from the inlet end.
Consultation with experts indicated that uncatalyzed hydrogenation of ethylene is not likely to proceed below 300 C. This advice was consistent with expectations, based on many years of ethylene plant operating experience.
For example, early units used high temperature front end acetylene converter catalysts. Feed to these reactors was raw gas containing ethylene and hydrogen. These reactors were sometimes operated at up to 240 C. inlet temperatures, with no known significant reaction in the preheat equipment.
Methanator temperatures had also increased in other units when the hydrogen stream contained ethylene, but a temperature rise upstream of the reactor had never been noted. Neither had a temperature rise been noted in driers regenerated with hydrogen and ethylene-containing streams.
This line of thinking led to the hypothesis that a catalyst must have been present. it was not considered likely that the methanation catalyst could have migrated out of the reactor through the top inlet and back through the 3in. feed line, double pipe 600-psig steam heater, and temperature control valve, to lodge near the inlet end of the feed/effluent exchanger shell.
Exxon arrived at the hypothesis that rust or reduced rust was the catalyst.
HYPOTHESIS SUPPORT
Reduced high-surface-area iron oxide is a known hydrogenation catalyst. Rust that forms on carbon steel equipment during plant construction can come off at start-up as micron-sized particles of iron oxide.
In this plant, abundant quantities of rust were seen in the equipment before and after start-up. The bulk of this material is usually flushed out in the precommissioning of new plants.
Prestart-up plans had called for most of the flushing to be done with compressed air, using the cracked gas compressor as the air source. Much of the equipment had to be flushed using an alternate, apparently less effective, procedure involving a repeated pressurization/depressurization cycle.
Significant quantities of rust, as well as other light debris, were found in certain equipment during the early operations following flushing. No accumulation of rust was found in the failed methanator exchanger following the incident.
This might have resulted from the effective flushing action of bursting the shell under 450 psi gas pressure.
Another unexpected phenomenon in the early operation of this plant points to unusual catalytic activity. Temperature rise in the main process gas 3A molecular sieve driers was seen during regeneration whenever the hydrogen-rich regeneration gas contained above-normal concentrations of ethylene.
During drier regeneration, pressure was roughly 50 psig, and normal inlet temperature 210-2301 C. Analyses of samples upstream and downstream of the drier regeneration system at the time of such a temperature rise showed a decrease in ethylene and hydrogen and an increase in ethane, across the system.
Instrumentation was installed, and procedures were instituted to operate safely despite this unexpected phenomenon.
The temperature rise could be controlled by reducing normal drier regeneration inlet temperature to a maximum of 190 C., and by lowering the temperature further if evidence of a temperature rise was detected. Other facilities and procedures were established to interrupt the regeneration and take further precautionary steps if the temperature rise continued.
Although no serious temperature increases were encountered, this phenomenon continued to be seen up to the first desiccant change, and has not been seen since.
Another Exxon plant had experience with regeneration of driers on its ethane cracking plant using a similar highhydrogen tail gas containing up to 10% ethylene, without such a temperature rise. These observations also fit the rust-catalyzed ethylene hydrogenation hypothesis, and no other explanation for the phenomenon could be found. These factors lent support to the hypothesis around the methanator exchanger incident.
LABORATORY PROGRAM
A short laboratory program was arranged to test the rust catalyst hypothesis. A novel laboratory approach to identifying the temperature at which an exothermic reaction initiates, using a Differential Scanning Calorimeter (DSC), was chosen for this work.
All experiments were run at atmospheric pressure in the DSC. The following potential catalysts were tested in the DSC: commercial nickel hydrogenation/methanation catalyst, reagent grade ferro-ferric (iron) oxide (Fe3O4), fine rust scraped from the inside of stored carbon steel pipes and vessels, and 3A molecular sieve desiccant.
The nickel oxide catalyst and the iron oxide and rust samples were reduced in the DSC with hydrogen. The nickel catalyst reduced over the temperature range 294330 C., while the reagent grade iron oxide reduced over the range 330-425 C. in the DSC test.
The DSC scan of the rust sample under hydrogen atmosphere showed that at least partial reduction occurred over a temperature range similar to that of the reagent grade iron oxide.
Tests were run in the DSC on the reduced-rust sample to determine its activity as both a methanation catalyst (CO and hydrogen were passed over the catalyst) and as an ethylene hydrogenation catalyst (ethylene and hydrogen were passed over the catalyst) at 350 C., and as an ethylene hydrogenation catalyst at temperatures down to 100 C.
Strong exotherms were detected for both reactions at 350 C. This indicates that, at this temperature, both reactions proceed rapidly over the reduced-rust catalyst. A strong ethylene hydrogenation exotherm was also found over the reduced-rust catalyst at 200 C., and a mild exotherm at 150 C.
But none was found at 100 C., indicating that exothermic hydrogenation initiates over the reduced-rust sample tested at between 100 and 150 C. at atmospheric pressure.
The CO/hydrogen mixture and the ethylene/hydrogen mixture were also tested in the DSC over 3A molecular sieve desiccant, at temperatures from 100 to 350 C. As expected, no evidence of exothermic reaction was detected in these tests.
Unexplained endotherms were detected at 100 C. in the tests with both ethylene and hydrogen over reduced rust and over molecular sieve. This may have been caused by some idiosyncracy of the experimental procedure (e.g., evaporation of water). No follow-up work is planned to explain this reaction.
The laboratory program confirms that reduced rust can act as a catalyst in the hydrogenation of ethylene. The lowest hydrogenation reaction initiation temperature found (100-150 C.) is more-or-less consistent with the location of the methanator feed/effluent exchanger failure.
The highest possible temperature reached in the exchanger without chemical reaction is the maximum heating medium temperature of about 250 C. The actual temperature at which the reaction initiated was probably closer to the feed inlet temperature of 5 C. than to the effluent inlet temperature of 250 C. The laboratory program found the minimum temperature for initiation of a rapid rust reduction at atmospheric pressure to be about 330 C. Exxon does not believe such a temperature was reached in the exchanger or the driers prior to experiencing ethylene hydrogenation. However, the lab experiment was of short duration.
Over the course of the several days prior to the ethylene carryover incident, the methanator facilities had been operating under hydrogen atmosphere at 32 bars pressure. It is believed that enough reduction had taken place at the normal operating conditions of the exchanger to render the rust at least somewhat catalytic. Once the hydrogenation reaction began, further reduction could proceed rapidly.
The lab tests also help explain the hydrogenation of ethylene in the drier regeneration gas at temperatures below 230 C. The lab tests show that over clean 3A sieve, no ethylene hydrogenation occurred at temperatures up to 350 C. However, the work suggests that if the sieves are contaminated with rust, ethylene hydrogenation can proceed at a much lower temperature.
CAUTION
The incidents reported here indicate that an accumulation of rust in new equipment is capable of catalyzing ethylene hydrogenation, and possibly other exothermic hydrogenations. These rust-catalyzed exothermic reactions have the potential to result in high-temperature excursions at locations in the process not anticipated in the design.
Although Exxon has not heard of a similar incident in any other ethylene plant, the possibility of rust-catalyzed reactions should be considered in the design and startup preparations of any process, such as the ethylene process, where the potential exists for commingling reactants that can hydrogenate exothermically.
Similar precautions should be considered in the restart of such a process if, during the shutdown, there has been an opportunity for significant rust accumulation.
Copyright 1991 Oil & Gas Journal. All Rights Reserved.
