SUPERHEATER PROBLEMS, THEIR CAUSES AND SOLUTIONS

John McMillanManager, EngineeringALSTOM Canada Inc., Ottawa, ON Canada

ABSTRACT

Superheaters in chemical recovery boilers operate in an extremely harsh environment. Mechanical problems candevelop which affects the integrity and reliability , and in extreme cases, the life of the superheater. This paper willreview some of the problems which have occurred with superheaters , discuss the causes of these problems, andoffer suggestions which will eliminate further damage, resolve the problem, and allow the user to better understandthe various operational parameters which affect the longevity and reliability of the superheater.

INTRODUCTION

Superheaters have been used in chemical recovery units for over 60 years. In order to perform their intendedfunction, they are suspended through the roof of the boiler and hang freely in an environment that is extremelyharsh. In this configuration they are exposed to a variety of internal and external loads and stresses which occur bothduring start up and operation. These stresses are a combination of static, thermal, cyclic, and transient in nature.Depending on the magnitude, frequency, and combination of these stresses, problems can occur which may affectthe reliability of the superheater . These problems can result in distortion of the assembly, cracks at attachments orseal bands, or tube leaks. The effect of the design of the welded and flexible attachments, the causes of in-plane andout-of-plane distortion of superheater assemblies, the damaging effects of excessive sootblower sway, and thestresses induced by various thermal loadings need to be understood in order to arrive at a cost effective solution tosuperheater problems.

SUPERHEATER ATTACHMENTS – TIE WELDS

Superheater platen assemblies consist of a number tangent parallel circuits which penetrate the roof of the unit andthen run vertically up and down inside the boiler and finally exiting through the roof again. Depending on theirconfiguration, they may penetrate the roof one or more times before they finally exit. While inside the furnace, thetangent tubes must be held in line to provide mechanical integrity to the assembly, to maintain proper alignment oftubes within the assembly, and to provide proper load transfer of the internal loops to the load carrying tubes whichpenetrate the roof. The weight of the assembly is supported by a seal band above the roof and is connected to thestructural steel by hanger rods. See Figure 1.

Over the years the attachment configuration and design has evolved and changed so that today, there are severaldifferent designs that currently exist in operation. Up until the mid 1970s the attachment of choice was the tie weld.These welds were 3 inches long, approximately 0.50 inches wide, and were spaced about 5 feet apart along theheight of the superheater.

Figure 2 shows a typical tie weld between tangent tubes. There were variations to this design whereby round orsquare bars were placed in between the tangent tubes so that there would be no weld directly between two tubes, aseach tube would be welded to the bar which filled the crevice between the tubes.

Experience with these tie welds varied. On some units they lasted the life of the superheater while on other units,tube leaks occurred. On the units that did exhibit leaks, it was discovered that cracks had developed at the toe of theweld and these cracks then propagated through the tube wall.

Figure 1 – Typical Platen Superheater Assembly

Figure 2 – Typical Tie Weld

Laboratory and field investigations into these failures resulted in two major contributing causes. 1) The weld metalwas massive in comparison to the thin walled tube, and during fabrication, as the weld metal cooled the compressive

stresses caused ovality in the tubing as well as high residual stresses. 2) During operation there is a constantlychanging heat flux in the boiler. High heat absorption raises the tie weld metal temperature causing it to expand.When the heat absorption is reduced, it will contract. This periodic change results in alternating stresses on the toe ofthe weld and, if high enough, will lead to fatigue failures.

In order to address these two problems, a major design change was made to the tie weld in the mid 1970s. The largetie weld was replaced with a single pass small weld which laboratory testing determined was stronger than the largeweld in resisting repeated flexing . The smaller weld had less mass which resulted in less shrinkage duringfabrication and therefore less residual stresses and ovality. Also, being smaller , the temperature differentials andfluctuations during operation were reduced. Figure 3 shows the comparison of the two weld styles while Figure 3Aand Figure 3B indicates a cross section of these two welds.

Figure 3 – Comparison of Large and Small Tie Weld

Figure 3A – Cross section of Large Tie Weld Figure 3B – Cross Section of Small Tie Weld

The newer smaller welds have resulted in improved reliability, as there have been less tube leaks experienced onunits that had the newer design. However, although this tie weld is more flexible than the previous massive weld, itis still a rigid attachment. As such, tie welded superheaters are subject to weld toe cracking due to fatigue. Thiscracking is caused by low cycle, thermal stress fatigue, due to stress concentration effects, residual stress, andmicrostructural changes in the heat affected zones. Figure 4 is an example of a crack at the toe of a tie weld. Weldtoe cracking due to thermal fatigue is a common phenomenon on attachments throughout the boiler. Where there is awelded attachment that is subjected to a cyclic loading, cracks can, and do, occur. All welded structures exhibitcracking in some form or fashion. The formation of cracks is not an indication of failure. On superheaters, weld toecracking by itself will not affect normal operation and should not be considered a sign of a problem or a defect.When weld toe cracking in tie welds occur, it is usually early in the life of the component, with most being

developed in the first year of operation. These shallow cracks, under normal circumstances, will tend to becomeblunt and inactive through the action of corrosion, the relieving of the residual stress by the cracking and thereduction in thermally induced stress with depth. See Figure 5 as an example of a dormant crack. As shown in theFigure, the end of the crack is rounded, or blunt. This crack will not propagate or result in a tube leak.

Figure 4 – Weld Toe Crack Figure 5 – Dormant Weld Toe Crack

In some cases the cracks do not become or remain blunt , but they will grow over time to the point where a tube leakoccurs. This happens when there is a cycling driving force high enough to create an active crack and the crack willgrow with each stress cycle until a leak occurs. Figure 6 is a photo of an active crack. This crack has a sharp end andwill grow slightly at each successive cyclic stress occurs. The determination of whether cracks are “old anddormant” or “active and growing” is made easier through periodic inspection and careful documentation. In terms ofinspection, knowing where to look is key to assessing the condition of the superheater. In platen superheaters, theplaten is comprised of parallel steam path circuits which operate at essentially the same temperature and pick upheat at approximately the same rate as they fold back and forth on themselves until the required heat pick up isachieved. In each of the intermesh tubes, the steam flow is in the same direction.

Figure 6 – Example of Active Crack

However, when the intermesh circuits reverse direction 1800, the result is two tubes of different temperatures beingattached to each other by the tie weld. This location is normally referred to as the weld between the outer loops.Figure 7 illustrates this concept. It is therefore at these locations where the stresses on the tie welds are the largest,and if cracks are to occur, these locations are where they would show up first.

Figure 7 – Attachment Location of Highest Stress

Routine inspections of tie welds should concentrate on the “ welds between the outer loops”. One of the preferredinspection techniques is the wet fluorescent magnetic particle testing. Careful documentation of all crack indicationsshould be recorded. In this fashion it is possible to compare the overall attachment condition from one inspection tosubsequent ones. The inspection only identifies crack “indications“. These indications may only be laps in welds,weld under-cuts, porosity, etc. Also, even if the indication is a crack, it may be an old, dormant crack which willnever grow to cause a leak. Thus, just as important as knowing where to look , is knowing how to interpret what isfound. There are various options available which does not include a wholesale replacement of the superheater.

If no leaks had occurred, a valid approach would be to document the locations of the crack indications and plan tore-inspect at a future outage in order to assess whether the cracks are growing in length, quantity, etc. A moreconservative approach would be to select a location that indicates a crack and cut out a sample so that it can bedetermined whether the indication is actually a crack, and if it is, whether it is blunted out or if it is active. If anactual leak had occurred at one of the welds between the outer loops, then it is recommended that other indicationsat the same location be repaired. This is done by grinding out the cracks with pencil grinders and repairing theground areas using a manual gas tungsten arc weld process (GTAW). This process allows for easy weld profiling atthe weld ends to reduce stress concentration and make fatigue crack initiation more difficult. These welds also havemore irregular and discontinuous side profiles to make crack propagation more difficult. The repair should then bere-inspected. An alternative repair is to grind off the entire weld, inspect the tube surface for cracks, and re-apply thetie weld. This technique is more time consuming and the decision as to which repair technique is used will dependon the number of indications, the outage time, overall costs, etc. Another option is to remove the tie weld and installanother type of attachment such as a hinge pin or a flexible tie. This is a good alternative to use for the attachmentbetween outer loops.

SUPERHEATER ATTACHMENTS – HINGE PINS

Another common attachment used to connect adjacent tubes in a platen superheater is called a hinge pin. Thisattachment resembles a door hinge and is comprised of short hollow barrels welded alternately on one tube with thenext one being welded to the adjacent tube. A solid rod or pin is inserted through the barrels to form the hinge.Figure 8 shows the hinge pin arrangement. Figure 8A is a cross section showing the alternate welding patternbetween the barrels and the adjacent tubes. The main advantage of the hinge pin over the tie weld is that it allowsdifferential movement between adjacent tubes caused by the differences in metal temperatures. It also allows slightrotational movements between the tubes. The result of these features is reduced stresses in the attachment weld,which make them less susceptible to weld toe cracking.

Figure 8 – Typical Hinge Pin Arrangement Figure 8A – Cross Section of Hinge Pin Arrangement

Hinge pins can, however, exhibit problems unique to their configuration. As their arrangement consists of a pinsliding inside a barrel, any fouling of the superheater can cause the hinge to seize , thereby negating the advantageof the hinge. When this happens, the hinge will not be able absorb the relative movements and will lock up the tubescausing distortion. The addition of extra rows of hinge pins can sometimes reduce the distortion to a tolerable level.

The barrel of the hinge pin will operate at a higher temperature than the tie weld because it is an extendedattachment of the tube. This will cause thermal stresses in the barrel to tube weld which may lead to weld cracking.Also, in high temperature regions of the boiler, the barrels may oxidize as they are not being cooled by the tube. Thepin itself is only attached by a weld to one of the barrels, or sometimes it is bent to prevent it from coming out ofthe hinge. It therefore receives little cooling and may be subject to wastage in the high temperature environment. Asthe hinge is designed to allow differential vertical movement between tubes, the longer the superheater assembly,the bigger the gaps need to be between the hinges. These gaps expose the pin to the gas temperature and wastage canoccur on these exposed sections. The danger in this is that the pin may fail and when this happens, the tubes withinthe assembly become disengaged from each other and will distort into the gas stream and may cause overheatingproblems, pluggage, or mechanical damage due to sootblower interference. Material upgrades that have a higheroxidation limit can resolve most of these problems. In some cases exotic materials may be required in regions ofextremely high gas temperatures. Figure 9 shows a photograph of pin wastage and reduced cross-sectional area inbetween the barrels.

Figure 9 – Pin Wastage Between Hinge Barrels

Being flexible and having gaps between the barrels can also result in bending or distortion of the pin itself . Oncethis happens, the hinge does not work and forces are set up in the hinge which will cause the tubes to distortbetween the attachments. This distortion, in turn, causes more pin bending, which causes more distortion and thiscycle continues until something cracks , breaks or operational problems prevent continued operation. The cause ofthe distortion and pin bending should be investigated in order to assess the most effective way to resolve theproblem.

IN-PLANE DISTORTION

Superheater distortion manifests itself in one of two ways, in-plane or out-of-plane. In-plane distortion is a distortionin the front to rear direction. The assembly takes on a shape like a banana having a convex side and a concave side.To understand the cause of this distortion, a simplified model can be used for illustration purposes. Consider avertical fixed cantilevered beam. If the beam is heated in such a fashion that the temperature varies linearly from oneside to the other, the beam will freely curve in space with no internal stresses. Once the heat is removed, the beamwill return to its vertical position. Figure 10 illustrates this concept.

Figure 10 – Temperature Distribution Linear with Depth

If the beam is heated in such a way that the temperature variation is not linear from one side to the other, the hotterside will expand but not freely. The cooler portion of the beam will restrain the hotter side, setting up compressivestresses on the hot side while the cooler side will be in tension. As the yield stress of a material is dependent on thematerial temperature, if the stresses are high enough, the hotter side will yield before the cooler side. In this case, thehot side will yield in compression and actually become a bit shorter than the cool side. Because of this compressive

Pin Wastage

yielding, once the beam returns to room temperature, the side that was hotter is now slightly shorter than the otherside and it will now be permanently bent in the direction opposite its curved shape when it was hot. Figure 11illustrates this concept.

Figure 11 – Temperature Varying Non Linearly with Depth

In a platen superheater assembly, as the steam flows through the tubes from the inlet to the outlet, it picks up heatfrom the gas flowing across the tubes. The metal temperature of the tube gradually increases from the inlet to theoutlet so that there is approximate linear variation of temperature from one side of the platen assembly to the other.During operation the superheater assembly would behave similar to the simplified beam in Figure 10. The convexside and concave side would be determined by the inlet and outlet tubes. When the boiler is out of service, thesuperheater will return to its vertical position.

If the temperature variation across the superheater becomes non-linear and the resulting internal thermal stresses arehigher than the hot yield strength of the tubing material, the superheater will still have a bowed shape duringoperation, but when the temperature is removed during a shutdown, a reverse curvature will occur due to thecompressive yielding on the hot side of the assembly following the beam concept in Figure 11. As the normaloperation of the superheater results in a linear temperature gradient, a non-linear temperature gradient is the result ofa condition existing in the superheater which prevents the normal heat pick up along the complete length of the tubefrom inlet to outlet. The cause of this non-linear gradient is most often associated with condensate blocking somecircuits during start up or water entering sections of the superheater during operation causing quenching and coolingof some of the circuits.

During a shutdown, the steam that is in the superheater tubing condenses and collects in the lower loops of theassembly. Another source of water in the superheater on a shutdown is from a hydro test that is done on thesuperheater as a result of normal maintenance procedures or following a repair on the superheater. The start upprocedure in the owners manual is intended to control the gradual heating of the superheater tubing while restrictingthe flue gas temperature in order that the water can be evaporated from inside the tubing in a controlled manner. Tooquick a start up will clear some circuits sooner than others so that steam flow will take the path of least resistanceand flow through the clear circuits while the water blocked tubes will have no flow. On the tubes with no coolingsteam flow, the metal temperature will approach the flue gas temperature thereby setting up a non-linear temperaturegradient from inlet to outlet as well as large gradient between adjacent tubes, one having cooling flow attached toone without any cooling. If this is the cause of the distortion, a review of the start up procedure should be done withparticular attention being given to when vent valves on the superheater headers are opened and closed. During startup the vent valves need to be open to provide a path for the steam to flow to provide the necessary cooling. Thesizing and location of these valves may need to be evaluated to ensure they can do the job they are intended to do.

Another source of water which can enter the superheater during operation is through the desuperheater spray valve.This will normally affect the high temperature superheater only. The water introduction can be due to 1) a broken ordefective desuperheater spray nozzle, 2) a spray water shutoff valve that does not close 100% when spray water is

not required, 3) a sensitive spray controller which hunts erratically and puts slugs of water into desuperheater inquantities that can not be evaporated by the steam, or 4) a spray water controller which leads or lags the steamtemperature feedback loop such that it is possible that under some circumstances the spray water can reduce tosteam temperature to saturation temperature.

A third source of water to the superheater is from carryover from the drum. This affects the first stage lowtemperature superheater. This is the result of upsets in the drum level control which causes water to be burpedthrough the steam dryers and into the superheater tubing.

The effect of in-plane distortion described above can be mechanical damage of the superheater assembly orassociated sootblowers. By distorting in-plane, the superheater assembly encroaches on the sootblower cavity.Depending on the location of the sootblower, interference may result between the sootblower lance and thesuperheater assembly. This interference can damage the superheater by rubbing against the leading superheater tubeto an extent that a tube leak results, or alternatively, the sootblower lance will be damaged, preventing it from beingretracted from the boiler or reinserted into the boiler. In some cases the distortion is high enough to prevent the useof the lower sootblowers and the unit will need to shutdown due to plugging or other operational problems.

One of the most common solutions is to cut a small section out of one side of the terminal tubes, raise that side andreweld the tubes. Although this will resolve the immediate problem, and allow sootblowing again, it will not resolvethe root cause of the distortion and the distortion will continue over time. This may be seen as a final resolution ifthe water ingress is a one time upset condition whose cause has been identified and appropriate actions taken toprevent its reoccurrence. Unless this is the case, the root cause of the problem should be identified and its solutionimplemented. Otherwise, the distortion will be an ongoing and re-occurring problem.

OUT-OF-PLANE DISTORTION

Out-of-plane distortion occurs when the tubes displace sideways out of the plane of the platen. This can happenbetween the top row of the attachments and the roof or the tubes can take a serpentine shape between levels ofattachments. In this case, a tube bows out of plane to the left side of the assembly, passes through the neutral point atthe attachment and then bows out of plane to the right side of the assembly. Figures 12 and 13 are photos of this typeof distortion. The serpentine shape suggests that the bowing is caused by compressive loading. As there are nomechanically applied forces that would create this compressive loading, thermal loading is the cause. The root causeof this thermal loading is the same as was identified as the causes of in-plane distortion, i.e. water entering the hightemperature superheater through the desuperheater, water carryover from the drum into the low temperaturesuperheater, condensate in the loops not being evaporated during a start up, or water from a hydrotest not beingcleared from all circuits during startup.

Figure 12 – Out-Of-Plane Distortion Figure 13–Out-Of-Plane Distortion of Lower Loops

Although out-of-plane distortion can occur on both tie welded and hinge pin attachments, it is more prevalent andpronounced with hinge pin connectors. When the axial travel is limited due to deformation, the pins are allowed tobend between barrels causing an axial restraint similar to tie welds. However, the hinge pins are still flexible inrotation and it is this rotation that allows the tubes to take on the serpentine shape.

Out-of-plane distortion can lead to pluggage in the gas pass between assemblies because the channels in between thesuperheater assemblies are partially occupied by the distorted tubes. Local overheating of the distorted tubes canoccur as the tubes will pick up more heat being completely exposed to the hot gases . In the platen arrangement, thetubes would normally pick up heat from both sides only. A third effect of distortion would be attachment weldcracks and possible tube leaks.

Once the tubes are permanently distorted out of plane it is almost impossible to force them back in line again.However, minor distortion can be mitigated by adding additional levels of attachments in between existing levels. Ifonly a few tubes exhibit severe distortion, they can be cut out and replaced with straight tubes. However, as statedwith in-plane distortion, the root cause of the problem should be investigated and resolved, otherwise, the distortionwill continue.

As a general distortion principle, both in-plane and out-of-plane distortion is caused by the introduction of, or thefailure to remove water from inside the superheater assemblies. This causes high thermal stresses which leads tobowing and deformation. The difference between tie weld and hinge pin attachments is how they respond to thisphenomenon . Tie welds respond as a unit causing the superheater assembly to act as a single structure while hingepins cause the assembly to respond as a collection of individual tubes.

SUPERHEATER SWAY

Superheaters are suspended through the roof of the boiler and are free to move relative to the boiler roof. By theirnature, all superheaters can sway back and forth during operation. During the sootblowing operation, the assembliesmove side to side as the blowers pass by each assembly on their insertion and retraction travel. This movement isnormally about 3 to 4 inches. However, if the assemblies sway in excess of about 8 inches problems can occur.These problems usually result in tube leaks at the roof elevation, at the seal band above the roof, or at the top levelof superheater attachments between the straight tube which goes through the roof, and the adjacent tube that formsthe top loop. See Figure 14 for location of sway failures. If failures at these location should occur, a swaymeasurement should be done to determine the extent of the movement and which sootblowers are responsible forexciting this sway. The sway test is relatively simple and involves attaching a horizontal rod to one of the assembliesand allowing it to protrude out of a local access door. A shield is placed over the door for safety reasons leaving onlya slot for the rod to freely cantilever from the assembly. During operation, the rod movement is measured as eachsootblower goes through its cleaning cycle.

Figure 14 – Typical Locations of Failures due to Sootblower Sway

Sometimes the resolution to the sway problem is relatively simple while at other times the sway can not be reducedand mechanical restraints are required to limit the damaging aspects of the movement. In one case the cause of theexcessive movement was traced to one particular sootblower, and on investigation, it was discovered that thesootblower nozzle was cracked. Once the nozzle was replaced, the sway was reduced to an acceptable level and thetube failures stopped. In other cases, varying the axial or rotational speed of the sootblower has reduced the sway.With the introduction of various high energy sootblower nozzles, the energy imparted onto the superheaterassemblies have increased which increases the sway. Reviewing the set pressures of the blowers and lowering themif possible can have the desired effect.

Should the sway still persist, mechanical restraints can be added to prevent the sway from causing problems. Steamcooled spacer tubes can be used to maintain the spacing and eliminate or reduce the sway. Bumper or satchel tubeshave also been used to minimize platen sway effectively. These types of restraints, however, restrict inspectionaccess as the assemblies can not be easily spread apart for inspection and maintenance.

THERMAL LOADING

In addition to the issues discussed above, there are other thermal loadings which affects the stresses and behavior ofthe superheater assemblies. In most cases these are secondary causes of problems but occasionally they are majorcontributors.

The superheater metal temperature is constantly cycling as the tubes become fouled and are then cleaned by thesootblower. The magnitude and rate of temperature decay and subsequent rapid rise of temperature varies based on

the fouling nature of the ash, the sootblower cycle, the sootblower sequence, the steam temperature controls, etc.This cycling nature results in the cycling of the desuperheater water control valve. Optimizing the sootblowercleaning sequence therefore can have a positive effect on the tube attachments stresses as well preventing largeswings in desuperheater spray water and the possibility of quenching the high temperature superheater.

Gas flow patterns in the upper furnace during start up or operation can affect the stresses in the assembly.Cleanliness variations throughout the assembly can result in the variations of stresses during sootblowing which caninduce additional cycling stresses. Selective plugging will change the local gas velocity in certain areas which, inturn, will change the metal temperature profile along the length of the tubes and this could lead to bowing of tubesor cracking of welds at attachments.

Temperature fluctuations can also be affected by variations in sootblower steam flow from one side of the unit to theother. Although the intent is to have equal steam flow going to each side of the unit, this is dependent on the sizingand layout of the sootblower piping.

CONCLUSION

Distortion, attachment failures, and tube failures are quite often associated directly or indirectly with

1) Condensate blocked tubes during start up2) Water from hydrotest not being evacuated during start up3) Water carryover from the drum4) Desuperheater spray water control issues5) Sootblower action6) A combination of the above.

The type of attachments used can influence the way the superheater behaves. It is important to determine the rootcause of the problem in order to incorporate any modifications or changes. Although this may require someinvestigative time, inspection and perhaps testing, it will result in a more comprehensive understanding of theproblem and will allow a more informed decision to be made with respect to the final resolution.