Furnace tube cracks on supercritical boilers 2

[athomas236]

I have been asked to advise on the the possible reasons and solutions for furnace tube cracking problem.

The sliding pressure boiler burns a severely slagging coal and supplies main steam to a 420MWe turbine at 3755psi/1055F. Although the turbine MCR is 420MWe it has an overload capability of 475MWe but typically runs at 460MWe during the day and reduces load to about 200MWe overnight to shed the slag off the furnace walls. The thickness of slag is typically 12 inches.

The furnace is spirally wound up to the furnace nose level where they are connected to transition headers after which the tubes are vertical.

The two tube failures have occurred on the one side wall after 30,000hours of operation. The failures are characterised by cracks starting from the outside of the tube and propagating inwards towards the bore. Furnace inspections have identified an area 4metres wide by 2metres high that is showing surface cracking. The cracks run at 90degrees to tube centre line.

The furnace has three rows of four burners in the front wall and two rows of four burners in the rear wall. The cracks have occurred just above the second row of burners.

The metallurgical report attibutes the cracks to thermal fatigue as a result of slagging/deslagging with evidence of overheating and oxide build up on the bore. The oxide is typically 65microns with a max of 85microns.
The overheating seems to occur overnight when tubes that are designed for 800F experience temperatures upto 950F.

Now my questions.

1. Why do tube temperatures increase when the load is reduced when I would normally expect them to decrease. There is a theory that the heat stored in the slag is being transferred to the steam when the load is reduced causing increased temperatures.

2. What is causing the cracks. To cause cracks at 90degrees to the tube centre line I would expect high longitudinal thermal stresses but cannot understand what is causing the longitudinal stresses to be greater than the circumferential.

Regards,

athomas236

 

[metengr]

Typically, thermal or corrosion fatigue cracks propagate (circumferential orientation) from longitudinal stresses versus hoop stresses (from internal pressure) in waterwall tubes. Sounds like what you have is thermal/corrosion fatigue crack propagation on the fireside tube surface, similar to what most people call "elephant hiding". You typically see families of these cracks that vary in length.

To answer your first question - how do you know that the waterwall tube metal temperature is increasing when you drop load? Have you measured it or just presuming?


If the cracking is occurring above the second row of burners and you have excessive oxide build-up on the tube ID surface with evidence of overheating, this sounds more like exposure to excessive heat flux during MCR operation of the boiler versus sliding pressure and dropping load at night. Also, with this much slagging your lower waterwalls are not picking up the heat resulting in having to fire harder and increasing exposure to flame impingement of waterwall tubes in the burner zones. Slag will loose heat quickly once you drop load so I doubt that heat retention would be the cause of excessive tube metal temperatures.


If you are slagging this much each day, I would consider installing water lances to keep those waterwall clean during the day. You might not have to drop load each night to shed slag.

 

[athomas236]

Metengr,

We are measuring the temperatures of the spiral wound tubes outside the furnace, before they enter the intermediate header at nose level and of the straight vertical tubes outside the furnace, before they enter the upper header at the top of the boiler.

As load is reduced from 460MWe to 180MWe the measured temps to the intermediate header drop from 390C to 325C whereas the temps to upper show a wide variation. At 460MWe the temps to the upper header are about 410C whereas at 28MWe they vary between 325C and 460C

Regards

athomas236

 

[metengr]

athomas236;
When you drop load and slide pressure, I could see where you would have some variation in temperature for the water/steam flow tube circuit going to the upper collection header versus the lower waterwall tube circuit that feeds the intermediate header.

However, what concerns me is the full load operation and the extent of slag build-up to the fireside tube surfaces during full load operation. From my experience with both super critical and subcritical boiler waterwall tube failures; it is high heat flux regions of the boiler waterwalls that will typically suffer from exposure to fireside thermal fatigue damage and tube metal overheating. Your deposit thickness on the waterwall tube ID surface is also the key to this conclusion. Have you had any under deposit corrosion?

 

[davefitz]

The root cause of the cracking on these units is that the furncae waterwall circuitry has a negative therohydrualic sensitivity charactireistic . That means, the tube which absorbs the most radiant heat will also draw the least amount of cooling water.

Basically , you have 1500 tubes operating in parrallel, with equal inlet pressure and equal outlet pressure, so if a tube absorbs more heat than average and the tube film friction exceeds the bouyancy effects, you will have this negative chracteristic. You can prove this by calcualtion,or you can add a lot of thermocouples on theoutlet tube stube and monitor it with a digital data acquisitio system.

When the furnace has excessive slag formation, the tube to tube heat absorption unbalance is much worse than the unit was designed for- in other words, current operations are in contradiction to a fundamental design assumption that was used by the boiler designers. The "worst tube" now has much less than the expected cooling water flow, and so when it passes thru the high heat flux burner zone, the inside film heat transfer coeficient deteriorates due to DNB ( below 3200 psia) or psuedo-DNB ( above 3200 psig). The outside of the tube may heat up to 1200 F, while an adjacent tube may be operating at a normal temp of 780 F.

These very tube to tube temperaure differentials cause a high thermal stress to be transferred across the membrane connecting the two tubes, and is one cause of cracking. In addition, the onset of DNB or psudo DNB will cause the presure drop in the tube to decrease to to a reduciton in film friction, so the flow to th at tube may suddenly increase to a normal flow rate, which reestablishes a high film boiling heat transfer coeficient, an this will cool the hot tube down to a lower value. This cycing of high tube to tube temperature differentials is the primary or root cause of the alligator cracking.

The high OD temps associated with this will also cause accelrated corrosion if there is sulphur in the fuel.

The boiler mfr will probably defend himself by saying the root casue is the factors that led to the high slaggin- either firing a off-spec coal or not maintaining the mill fineness within spec.

 

[davefitz]

I forgot to add that another thermal stress, perhaps the primary thermal stress, is the cold side to hot side tube temperature differential on the worst tube.

If the tube is operating with the bulk fluid slightly below the "psuedo critical point" ( ie point of max heat capacity when above 3200 psia, about 720 F), a high heat flux can casue the film properties close to the ID on the hot side to deteriorate, as the viscosity and conductivey drastically decreases as the film temp increases from 720 F to above 760 F. This causes the hot side of the tube ( facing the furnace flame) to be heated to above 900 F while the cold side of the tube is held close to 720 F.

This causes the temperature differential between the hot and cold side of the tube to exceed typical fatigue limits of 200 + F. This high a front to back DT leads to a high bending moment about a horizontal axis, which causes the hot side to yield in compression. When the overheat has passes and normal tube temperatures are restored, the yielded hot side will then be exposed to a very high residual tensile stress, which eventually leads to the circumferential cracks they call alligator cracks.

 

[davefitz]

The fix for your specific problem , in order of cost, is as follows:
a) switch to a fuel with the origicnal spec slaging properties.
b) maintain the mills so the coal fineness is within spec. This is crucial to minimizing the burner to burner coal flow unbalance, which could lead to substochiomentic operation of some burners, and the resultant impact of unburned char particles on the waterwalls.
c) upgrade the wall blowers to water deslaggers, but must rigorously adhere to the Diamond sootblower guidelines on operation of water deslaggers to avoid damage to the spiral wound furnace waterwall
d)At low loads, bias the burner pattern centerline to the top burner level. This reduces the effect of a negative thermohydraulic sensitivity characteristic.(Ie take bottom row of burners out first)
e) add more inlet orificing pressure loss at the inlet of the waterwalls
f) convert waterwalls to the Siemens KWU vertical rifled tube design with a positive thermohydraulic sensitivity characteristic.

 

[metengr]

davefitz;
Generally agree with your recommendation list except the fuel switch is not the least cost option by any means. You have transportation issues and environmental concerns. I would not rank fuel switch as the least cost.

 

[athomas236]

metengr and davefitz

Firstly, thank you both for you assistance on this matter and apologies for not responding earlier. I was delayed by the clients requirement for a preliminary report by no later than 11pm Sunday UK time. I just made it with about 10mins to spare.

To hold the client while I think some more I have said the following based on the limited information available to me.

1. Thermal fatigue in the most likely cause of the failure resulting from the thermal cycling caused by slag being detached combined with excessive tube temperatures. EPRI report Water-touched tubes, chapter 19 seems to support this view.

2. The internal oxide thickness seems excessive after only 30,000 operating hours and a survey of the oxide thicknesses could be beneficial and this could be done without taking tbe samples.

3. The thermal inertia of the slag on the tubes doesnot explain the high steam temperatures at reduced load.

4. Excessive tube temperatures could be associated wth the reduction in internal heat transfer coefficient combined with local high fluxes and more information is required if this is to be investigated further.

I have asked for more drawings and operational, coal and ash data and details of any attempts that have been made so far to reduce slagging.

At the moment I am trying to understand the negative characteristic and deterioration in heat transfer coefficient mentioned by davefitz but there is very little published data that is readily available.

Regards

athomas236

 

[davefitz]

The negative sensitivity characteristic was initially discussed in the 1930's in Germany by Ledinegg, and also later discussed in the USA in the 1950's when supercritical steam cycles were initially discussed for liquid metal cooled reactors. Today there are a few commercial discussions ,mostly by boiler mfrs or by Siemens KWU - see their papers by J Franke and Kiefer, Kohler and Hein ( Int'l j of multiphase flow).

The deterioration of inside film heat transfer was discussed in the 1960's by Russians and in the 1970's some good papers were publised by B+W. The topic was mostly kept quiet as it represented some "dirty linen" by the boiler mfr's.

See allso books by Sadik Kakac, Older papers circa 1958 by Hyman in proc AiChe, and the best paper on modern techniques for theral hydraulic circuit senistivity is by F Thelen, VGB Kraftwerktechniks, V61 no 5 may 81 pp357-367, and their is also a Dounreay translation

 

[davefitz]

High waterwall circuit outlet steam temperatures at part load can be caused by a number of things.

If the average outlet temp is higher at low loads, it can be due to either (a) the programmed superheater spray water flow was erroneously set at a fixed flow rate instead of a fixed percentage of current load, or (b) the burner pattern was incorrectly biased to the lower elevations . It is only proper to bias the burners downward at part load if the furnace waterwall has a positive sensitivity characteristic,as with natural circulation drum boilers- it is not a good practice on once thru steam generators.

If the average waterwall outlet temp is OK but the "worst tube" temperature gets worse at low loads, it is probably because of one of the following: (a) the circuit inlet orificing was originally determined to correct the negative sensitivity characteristic at full load, but the effect of inlet orificing drops as load drops (b) the fluid temp leaving the economizer and entering the waterwall is too high .

 

[davefitz]

athomas:
Some newer commercial papers that provide an introduction on the topic of thermal hyraulic sensitivity of the waterwall of a supercritical OTU are:

"Refurbishment of the Yaomeng Power Plant" by Mitsui Babcok for DTI, cleaner fossil fuel programme, bulletin BPB005, see fig 13

"Technology on the March: Steady Progressin Supercritical Once-through Technology", by Steve Goidich, Foster Wheeler , presented at the July 2001 Coal Gen conference.

 

[athomas236]

davefitz,

I have ordered the Thelen VGB paper and downloaded the Yaomeng publication. Will continue searhing for the others.

I remember Ledinegg from my days in the early 70's when I was a boiler engineer. He had a big impact on the LaMont Boiler Association in Lodon which resulted in orifies being installed in the tube inlets of their standard designs.

Thanks again

athomas236

 

[davefitz]

The paper by Steve Goidich is a good summary and introduction.

Some older papers that also give a simplified engineering approach are :
"Flow distribution among parallel heated channels" A R Gruber, S C Hyman, Chem Eng Prog v2No2Jun 56 pp199-205

"Control of flow distribution by mixing headers", S C Hyman, A R Gruber, L Joseph Chem Eng Prog v4 no 1 mar 58 pp 33-36

Technically, the low load overheat at the burner zone is not DNB- it would be "dryout chf" if the pressure is below 3200 psia since the steam quality at that zone is over 30% SBW. The prediction of heat flux vs mass flow ,P, SBW would be as per Kon'Kov. See "Heat transfer in an evaporator tube with circumferentially non-uniform heating" K Becker, A Enerholm, L Sardh, W Kohle, W Kastner, W Kratzer IN J multiphase flow v 14 no 5 pp 575-586 1988.

 

[athomas236]

Have ordered the Gruber and Becker papers and should have all the papers by Friday this week.

At the moment I am continuing with my investigations using Bishops correlation which does indicate a deterioration in heat transfer coeff at 380C.

I will probably be on more comfortable ground when I come to the sub critical part load cases. It will just be a few hours in the loft lookin out my notes from the 70's.

Thanks again

athomas236

 

[davefitz]

see http//

http://www.fosterwheeler.fi/publications/tech_papers/powergen/pdfs/TechontheMarch.pdf

for steve goidich's paper

 

[athomas236]

davefitz,

I have downloaded steve's paper.

 

[davefitz]

AThohmas:
I haven't heard how you did with the cracking study.

Another instigator of local waterwall tube overhreat on supercritical units is related to the large difference in response times between the control fo feedwater flow and the control of coal firing rate. See the book that is edited by S. Kakac " Boilers, evaporators and condensers" for some examples of this phenomenon.

Normally a once thru boiler controls feedwater in the quasi-steady state by maintaining a constant ratio of (firing rate/ feedwater flow). Upsets in final steam temperature can be addressed to minor extants by spraying at the interstage desuperheater, but this spray water is meant to be set in the quasi-steady state at about a 0.5% spray flow, and this is done by perturbing the ratio of FR/FW.

When there is a input for an increase in steam flow or megawatts, the demand for coal flow is increased, and the feedwater is increased in direct proportion to that demand in FR to keep the ratio of FR/FW constant. Unfortunately, the FW process has a very fast response, with a time constant of about 3 sec ( merely needs to open the FW control valve, with some lag if a steam turbine driven BFP is used), while the FR response is much slower.

The coal mills have a much longer reponse time , with a time constant of about 1-2 minute. This is because initially there is a step increase in mill table air, which will elutriate a significant increase in coal to the burners, but once the table coal fines inventory is depleted, it takes perhaps 1-2 mintues to refill the table with fines for the new steady date coal grinding rate. This hysteresis in FR response leads to an upset in enthalpy at the furnace waterwall outlet. During load decreases, this can lead to a temporary enthalpy overheat excursion at this location, sufficient to initiate a psuedo DNB event at this location. The mechanism which forms the alligator cracks is due to the psuedo DNB event , as outlined previously.

Siemens has proposed ( and patented) a concept where the enthalpy is measured at this location using gamm ray detectors, but there are other mfr's of these devices today.The realtime indication of outlet avg enthalpy can be used as a trim to feedforward an adjustment of the FR/FW ration to avoid these upsets.

 

[athomas236]

Still thinking about it

Its 5pm UK time on a saturday afternoon been down my village pub

I have till 11 pm tomorrow to make my report, will think of some thing

athomas236

 

[eyec]

athomas236
sorry for the late post but a little operational history would seem in order.

have you always had this excessive slagging situation?

if no, then what changed?

all of the discussion around DNB seem to address the effects of the problem not the root cause.

davefitz' last post seems to be getting nearer the root cause.

slagging tends to result from incomplete combustion of the fuel.

what is the position of the fireball in relation to the high heat areas exhibiting the elephant skin?

what examination of the fuel to air mix have you performed at the various load ranges?

in my opinion, the FR/FW ratio and fuel to air balance may be the root cause.

anyway, good luck and please keep us informed.