Failure of Waste-Heat Boiler afterDebottlenecking Process Air Compressor

A few months after raising the efficiency of the process air compressor, one ofthe tubes in the waste-heat boiler downstream the secondary reformer failed.

One possible reason was the malfunction of the water circulation inthe tubes of the boiler.

Edgar Lebold, Joerg Reininghaus, and Roland SchoberBASF AG, D-6700 Ludwigshafen, Germany

In November 1991 BASF's AmmoniaPlant No, 3 in Ludwigshafen had an emergencyshutdown caused by an untypical failure of thewaste-heat boiler downstream of the secondaryreformer. The high pressure steam generatingsystem had a sudden loss of water due to aruptured tube in one of the four boilers.

After cooling down the system and openingthe gasline the following investigation on thedamaged boiler reported one hole, severalsmall leaks and erosion in one of the tubesand also a small leak in one similar tube in theneighbourhood. Even the part of the tuberenewed Q weeks before had been erodedagain.

The whole boiler was replaced by a newone of the same type in a 4-week operation: 27high pressure tubes and 3 gaslines had to berenewed.

The analysis of the causes for thedamage showed that due to the unsuitableconstruction of the boiler changes inrunning the equipment were necessary toavoid failures in the future.

THE PLANT

BASF's Ammonia Plant No. 3 inLudwigshafen is a single train plant with adaily production of 1300 mtpd NH3. Theplant was engineered and constructed byBASF end of the sixties with a designcapacity of 1150 mtpd of NH3. The processsteps are: steam reforming of natural gas,secondary reformer with waste-heatboilers, high temperature and low-temperature shift conversion, adsorption ofC02 by BASF's activated MDEA-process,methanation, compression and a synloopoperating at 300 bar with a 4-bed-quench-reactor (Fig. 1).

146

The plant started up with production inJanuary 1971. The capacity was limited by theprocess-air compressor. To exceed the designcapacity the additional use of expensivenitrogen from the central supply net wasnecessary. This nitrogen was compressed byan electric driven reciprocating compressor andadded to the process gas at the methanation.

THE PROCESS-AIR COMPRESSOR

The process-air compressor built by Borsigis a 4 stage centrifugal compressor with twocasings: having low and medium pressurestages in one, the high pressure stage in theother casing. The compressor is driven by a 16bar steam-turbine. Original design capacity was41.000 mVh air with a steam consumption of39 mtph (fig. 2).

After renewing internals, roters and statorsthe capacity of the compressor was increasedup to 45.000 mVh with an additional steam-consumption of only 1mtph. The quantity ofmore process air was sufficient to produceammonia up to 1300 mtpd without the use ofnitrogen from the central supply net.

The increased process air flow also resultedin a higher load of the secondary reformer andthe downstream equipment especially of thewaste heat system.

THE WASTE-HEAT BOILER

The damaged waste-heat boiler is part of asystem comprising four boilers with naturalcirculation: two waste-heat boilers downstreamthe secondary reformer, an auxiliary boiler firedwith natural gas and linked to the fluegas ductof the primary reformer and a process gasboiler downstream the high temperatur shiftconverter. The boilers share one commonsteam drum and produce together up to 250mtph of 105 bar-steam (fig. 3).

The damaged waste-heat boiler is themost thermally stressed apparatus in thesystem (gas inlet temperature up to960 °C, outlet temperature about 440 °C).It is a watertube boiler comprising threemain components: the pressure vessel,rectangular boxes built by fin tubes and avertical tube bundle, which is inserted intothe boxes (fig. 4).

The fin tubes form three passes for theprocess gas: one hotter pass inside andtwo cooler passes outside. The tubebundle is inserted from the top: 8 harpsinto the cooler pass, 12 into the hotterpass. Every harp consists of a verticalbayonet type header and 14 riser tubes forthe steam water mixture (fig. 5).

The riser tubes are connected byhorizontal headers leading the steamwater mixture to the steam drum.Fin tubes and tube bundle are fed withboiling water from the steam drum undernatural circulation conditions (fig 6).

Due to this construction, it isimpossible to empty the waste-heat boiler.Uprising concentrations of salts can onlybe avoided by draining the steam drum.Also solid components also can onlyescape via the steam drum. The maintechnical datas are presented in table 1.

FORMER FAILURES OF THE WASTE-HEAT BOILER

Since start up of production in 1971there were 8 failures of this waste-heatboiler. The failures are listed in table 2 andthe location is shown in figure 7. Threedamages may be regarded as singleevents:

1975 hole in a cover plate of a steamheader,

1980 defect insulation in the gas inletzone,

1988 ruptured U-tube at a header ofthe fin tubes.

147

All other damages were situated on theslope or vertical riser tubes. In 1988 the numberof problems with damaged riser tubes grew up.So within four weeks the tube bundle wasdismantled and the hotter section wascompletely renewed (flg. 8).

It was worked out that the quality of theboiler feed water and missing addition ofhydrazine were reasonable causes for thedamages. The bottom covers of the verticalheaders contained a handful corrosionproducts. These might have caused amaldistribution of the boiler water, a localthermal overload and even a dryout of the risertubes.

Table 1: Waste-heat boiler design data ondoperating conditions.

Table 2: Failures of the riser tubes.

design data:steam generating capacity 130 t/hheat duty 189 GJ/hheating surface 440 m2

material of tubesand headers

13 CrMo 44ASTM A387-65B

design metal temperature:water side 350/375 °Cgas side 350 °C

design pressure:water side 122 bar ggas side 34 bar g

operating conditions:steam pressure 105 barggas pressure 29,5 bar ggas temperature

inlet 990 'Coutlet 420-500 *C

Since October 1988 the pH-value of theboiler water was set between 9.3 and 9.7adding sodium hydroxide. Hydrazine was dosedpermanently and the blowdown of boiler waterfrom the steam drum was increased. As ameasure of precaution a spare waste-heatboiler was ordered immediately. It was used inthe repair action in 1991.

1. May 1985

2. March 1988

3. Sept. 1988

4. Sept. 1991

5. Nov. 1991

leak

hole (2 mmdiameter)oval hole (13mm by 15 mm)small hole(1mm dia-

meter)small crack

leak

hole (80 mm by40 mm)2 leaks

FAILURES iN 1991

The damages in 1991 were similar tothose in 1988: During a shut down inSeptember a small leak at a sloping risertube was noticed. The leak was on theoffstream side of the tube and caused onlya small loss of water. The damaged part ofthe tube was cut out and a new piece waswelded in.

Six weeks later the plant tripped due tothe sudden loss of water in the steamgenerating system. The steam blew downthrough the waste-heat boilers, the hightemperature shift converter and ventedthrough the flare. We were lucky that thecatalyst pellets of the HT shift converterdid not entrain with the steam as ithappend in 1988. Because of wetting thecatalyst heavily and the resultingmechanical stress to the pellets the HTshift catalyst hat do be replaced.

After cooling down and venting thesystem we could inspect the waste-heatboiler: The riser tube repaired six weeksago was ruptured in the vertical sectiondownstream the renewed part. Inneighbourhood to this part justdownstream the weld, there was a smallleak.

148

The investigations on the renewed tubereported heavy erosion in the six-week old part.Another riser tube in the neighbourhood insame position had a leak in the slope section.Both leaks and the zone of erosion weresituated on the offstream side of the slopingriser tube. The rupture was situated in thevertical section along the ruling line parallel tothe gas flow (fig. 9).

INVESTIGATION ON MATERIALS

The metallographic examinations reportedsimilar causes of the damages in 1988 and1991: In the zones of erosion and leaks theprotective magnetite layer was irregularylaminated like a system of independantsandwich-like layers (fig. 10).

We assume: The high gas inlet temperatureand the increased gas flow due to therevamped process-air compressor caused alocal thermal overload of the waste-heat boiler.Especially the riser tubes in the gas inlet zonetended to be overloaded. Spots of disturbedmagnetite layer may have caused a high rate ofbubble formation and finally a local dryout ofthe tube wall resulted. The magnetite layerburst because of the different thermalexpansion of steel and magnetite layer. Themagnetite layer was formed again byconsuming tube material. The thickness of theporous, disturbed magnetite layer grew andheat transport went worse. The result was afurther increase of temperature at the tube walland a burst of the magnetite layer. Thepermanent repetition of theis process causedthe erosive pull down of the tube wall untilfinally a leak or rupture occured. We noticeddamages of only two tubes, in both casessituated on the lowest tube of the harps in thehot gas inlet of the waste-heat boiler.

CAUSES OF THE DAMAGES

The boiler feed water of the AmmoniaPlant No. 3 is a mixture of demineralizedwater from the supply net and reused,untreated condensate of the turbines in theplant.

A part of BASF's demineralized wateris produced out of water of the river Rhine.It may contain small residues ofchlorinated hydrocarbons which will becracked under the conditions of the waste-heat system. The aggressivity of the insitu formed hydrochloric acid will beneutralized by higher pH-values. Thereforeit is necessary to add sodium hydroxideinspite of the disadvantage of speeding upcorrosion and erosion in the case of dryoutby settling in the magnetite layer andmaking it porous. Investigations on therecycled condensâtes report that theyimprove the quality of boiler feed water.

When in 1988 the damages hadaccumulated, the conditioning of the boilerwater was thought over and analysis wasintensified:

- pH-value set between 9.3-9.7 byaddition of sodium hydroxide,

- Addition of hydrazine as an inhibitor,

- Increase of the blowdown of waterfrom the steam drum,

- Weekly analysis of the total steamgenerating system.

Besides the quality of the boiler-waterthere are other important facts of influence.The rupture of the vertical riser tubeindicates a heavily disturbed flow of thesteam water mixture. Formation of bubblesmay stop the natural draft circulation whichis induced only by a small pressuredifference (fig. 3, 5).

It is very dangerous if counter-flowingsteam bubbles hinder the water supply tothe lower, hottest parts of the verticalheaders.

149

Like steam bubbles, corrosion products andsludge in the bottom covers of the verticalheaders prevent the sufficient cooling of thelower riser tubes. Because there is nopossibility to drain the vertical headers,corrosion products are accumulated for yearson the lowest part (cover) from where the lowerevaporator tube is supplied. In 1988 x-ray testsshowed a lot of sludge in the covers, which wasthen removed. In September 1991 video-checksreported only a small amount of corrosionproducts which mainly escaped in November1991 because of the rupture.

These special fluiddynamics are a problemat high thermal load of the waste-heat boiler. Acheck of the process parameters indicates athermal overload (steam production) of 10 %over design. Before the damges in 1988 and1991 the waste-heat boiler had been runningdose to the allowed maximum of gas inlettemperature of 990 *C. Finally tine revamp ofthe process-air compressor increased the heatflux furthermore.

ACTIONS

Even before installing the new waste-heatboiler it was cleaned because of theproblematic fluiddynamic situation. Althoughstored under conservation conditions (nitrogen)the tubes of the new waste-heat boilercontained a lot of corrosion products in the risertubes. These were removed by a complicatedprocedure of steam blowing for several times.The success of this cleaning procedures wascontrolled by video-checks. To improve thequality of the magnetite layer we started uprunning the waste-heat boiler according to aspecial pH-value programm.

Furthermore we used the ASPEN simulationprogramm to set up economically andtechnically reasonable process parameters forthe waste-heat boiler. The process-aircompressor is still running under the conditionswe got after teh revamp. In order to avoid toohigh gas inlet temperatures, we run themethane concentration downstream thesecondary reformer above a critical level.

By these measures we are sure, thatthe thermal load of the waste-heat boilerremains ai a technically acceptable lowvalue, as we had run the plant before thedebottlenecking.

At any shutdown of our plant the risertubes and their headers will be checked.

All our experiences show that water-tube type waste-heat boilers are verysensitive to thermal load. Especially whenthey have no drains and when they arerun only by the small forces of naturalcirculation.

CONCLUSIONS

There was an untypical failure of thewaste-heat boiler after revamping theprocess-air compressor and increasing thecapacity of the Ammoniak Plant No. 3.Due to the construction of the boiler incertain tubes the two-phase-flow of steamand water was disturbed at high energyload conditions. The tubes were eroded.Such a damage was not predictable whenwe planned the revamp of the process-aircompressor.

Since the last repair in November 1991no signs of tube failures were detected inthe waste-heat boiler.

150

DISCUSSION

T. S. Hariharan, F er til, Abu Dhabi: I understandthat the high heat load is the probable cause for thisfailure. What do you consider to be an acceptablelevel for heat load?Reininghaus: I do not have the numbers at present.Hariharan: When you ordered a spare boiler didyou do any modifications to reduce this heat flux?Reininghaus: No. It is an identical boiler.Hariharan:. We have a 1,000 tpd ammonia plant,and the fired tube boiler has suffered three failures.

We have come to the same conclusion as you thatthe high heat flux is the main reason. We are nowconsidering a lower heat rated modification.P. J. Nightingale, ICI, England: We had a similarexperience. Did you monitor the level of chloridesin your boiler and change your operating patternaccordingly?Reininghaus: It is difficult to be exact. Levels ofabout 1 ppm could have existed in the drum.

151

Figure I. BASF's Ammonia Plant No. 3.

Figure 2. Process-air-compressor.

Û

Steamdrum

ÛFeed-water

fromSecondary ! ym

Reformer

Auxiliaryboiler

Primary Secondarywaste-heat boiler

Waste-heatboiler

Figure 3. High pressure steam generatingsystem.

steam/water

boiler feedwaier

Cf sïeam-/vwaler

'S'ïureir 1975

'ai-ure in 19EO

<Lf

uboiler feedwater

gas from secondary reformer

Figure 4, Waste-heat boiler.

boiler feedwaier

Josteam/water ^— -—*•,-*-•] "Z"

!

3 faujrein 19B8-A-i boiler leedwaier

gas from secondary reformer

Figure 5. Waste-heat boiler.

152

steam/water

vertical header

Figure 6. Scheme of a tube harp.

Figure 8. Renewed lower part of inner risertubes.

w0 «%1

(D

@

(S)

©

Figure 7. Location of failures of the risertubes.

i

r?

—

Lt7

ni

»a

-

ks

- —

Ruptured tube

j f

Gas f low

Figure 9. Ruptured riser tube.

153

Roland Schober

Figure 10. Irregular magnetite layer.

Edgar Lebold Joerg Reininghaus

154