Colloidal Silica Problem??? 2

[exnavynuke]

Hi all,

We are experiencing a vicious and repetitious silica problem at this relatively new power plant. It has been an intermittent ongoing problem.

Plant design: 2x1 GE gas turbines into one GE steamer. HRSGs are triple pressure reheat. Demineralizer is Cation, anion and mixed bed. No pre-filtration of the raw water, which is supplied by the local water district from wells.

Here's the problem: If we run base load or less, silica is under control. Operating HP drum pressure is 1100 psi or so. Whenever we fire duct burners to increase power output, HP drum pressure goes to 1800 psi, and suddenly silica starts appearing. It can get bad, like up to 50 ppb in the condensate.

We aren't seeing an increase in hardness or conductivity in the drums, so we don't really suspect a condenser tube leak. Besides, we had a contractor perform a helium on-line leak test recently.

It's my understanding that subsurface waters don't generally have colloidal silica, but based on what we've eliminated so far, it looks like we're down to that.

The suspicion is the colloidal is in the boilers unreacted, until the duct burners break it down into reacive silica when we go to high power.

Question: Does anybody know what temp/press causes significant breakdown from colloidal to reactive? This would really help confirm what we suspect. I've hunted all over the internet for this and can find nothing.

 

[hacksaw]

what is your target silcia level in the bfw.

 

[exnavynuke]

Less than 10 ppb in the condensate and BFW. Less than 400 ppb in the IP drum and less than 300 ppb in the HP drum.

 

[hacksaw]

Those are fairly normal ranges for steady-state operation at either 1100 or 1800 psi.

Suspect that at low rate you are getting silica deposition that is then vaporizing as you go to full pressure operation. To eliminate the problem you'll need additional equipment in the lineup. You'll have to set down with the boiler manufacturer on this one. It would be a simple problem if it was just steam cleaning involved.

Good luck,

 

[bimr]

Can't help with the silica question. I will add that most power plants are using RO/DI. RO/DI would provide higher water quality than you presently have.

I agree with hacksaw's silica theory.

 

[cub3bead]

1. Ground water may have colloidal silica in it and it can be significant. It depend on the formation geology that the water is produced from.

2. The operating pressures you are at is more than sufficient to convert the colloidal silica to reactive silica. I have seen the conversion occur below 600 psig.

3. If the problem is silica hideout, putting a RO in the treatment line won't help. The plant is relatively new were the HRSGs chemically cleaned prior to startup? If so what cleaning agents werre used? Hideout becomes a problem in dirty steam generators.

4. What condensate and boiler chemicals are you using?

5. Have you been monitoring the makeup system for colloidal silica in the effluent?

6. What is the source of the desuperheating water for the mainsteam and reheat?

 

[exnavynuke]

Hi and thanks for all the replies...

Our demin water tank has never been 'hit' by breakthrough - never tested above 5 ppb on the makeup, so that would not seem to be the source. It's gotta be colloidal or else left-over from construction.

To cub3bead:
1. We've begun performing tests on the influent water for colloidal silica, but have no results as yet. Also the water supplier routinely takes wells out of service for maintenance, and places others in service, without notifying us. So even the samples we are currently taking may not show anything colloidal.

2. I thought 1100 psig would be more than enough to convert colloidal into reactive silica, but it seems to be more of an *equillibrium* between the two forms. If 600 psig is enough to convert silica, one would think that we would always have a silica problem then at 1100 psig. Yet we have excellent control of silica with minor BD at baseload. When we go into power aug, look out!

3. Agreed that pre-treatment won't help a boiler that is plated with contaminants The plant is about 1-1/2 yrs old now, and yes the HRSGs were cleaned. Not sure what chemical was used, although it may have been an EDTA or other mild acid. Demin trailers were used to supply the wash and rinse water during commissioning, and afterwards, each section was thoroughly bottom blown under light steaming conditions. The drain lines are not rated for full working pressure, so that cannot be done.

4. Nalco Eliminox oxygen scavenger, BT-4000 drum coordinated phosphate/ph control, 352 neutralizing amine for the pre-boiler and steam ph

5. See #1

6. Steam attemporation is from BFW pump IP discharge for the reheat steam, and HP discharge for the Main (HP) steam. The BFP takes suction from the LP drum.

Of interest is that the silica first shows up elevated in the condensate pump discharge and LP drum, then cycles up in the HP and IP drums. I would have thought that if there were a colloidal problem (or a plating problem) it would first be observed in the HP section and show up high there, then move elsewhere via the steam... weird.

Thanks again for all the assistance, and any further assistance!

 

[bimr]

If you have silica in your system, it probably got in there in your makeup water.

Are you sure that you are measuring for total silica in the makeup? The automated online instruments probably only determine the concentration of reactive silica. The measurement of total silica is generally considered an off-line procedure; in most cases, a water sample is transported to a lab for analysis.

Colloidal silica has created problems for water treatment because of its stability as an unionized compound, making it difficult to remove using ion exchange processes.

Silica is also found at the lower end of selectivity for anion resins, creating a scenario where silica breakthrough is one of the first to occur. As a result, silica can be effectively removed only if the ion exchange resins are completely and properly regenerated.

With the introduction of RO as a pretreatment process, the colloidal silica can be removed very effectively simply by virtue of the filtration capabilities of a typical RO membrane. In fact, if one uses a conventional definition of colloidal silica as that which measures greater than 50 angstroms, even ultrafiltration with a molecular-weight cutoff (MWCO) of 100,000 daltons is effective at removing colloidal silica. The RO also helps the ion-exchange process by alleviating the overall silica loading on the resin.

I might add that some pretreatment is prudent on your makeup water supply. You are trying to make ultrapure water and there is typically a lot of crap in a municipal supply.

 

[exnavynuke]

UPDATE: This problem has been identified. But first...

Since the original suspected problem was colloidal silica getting through the demin train, the plant has installed temporary Multimedia filters and coagulant injection upstream. Then a trailer-mounted two-stage RO was brought in downstream of the MMF, but upstream of the demin train.

The entire system was run for two weeks, with no change in condensate or drum silica levels.

This week one circ water pump had to be removed for mechanical repairs, and the plant was de-rated 100 MW. So out of desperation we isolated 1/2 of the main condenser and drained the waterbox. Nothing changed over 24 hours. We refilled it, isolated and drained the other half of the condenser waterbox. Voila! Drum silica levels were 1/2 their previous level after 2 hours (from 1000 ppb to 500 ppb).

Lessons learned:
1. Try the easy/inexpensive things first.
2. Don't rely too much on on-line leak testing.
3. Don't assume that a condenser tube leak is always accompanied by high conductivity and hardness levels. In this case it wasn't.

 

[cub3bead]

exnavynuke,

Congrats on finding the problem. Shame on me for not mentioning a condenser leak as a possibility, guess the tunnel vision was in full mode that day.

Do you monitor cation conductivity out of the hotwell or just specific conductivity? Cation conductivity will respond a bit quicker but you should know that from your Navy days.

 

[exnavynuke]

Thanks cub3bead...

It should not have taken so long to find a minor condenser tube leak, but everyone here was focused on the colloidal suspect as well.

We've spent several thousand dollars on regenerant chemicals, demin trailers, the MMF and RO trailers. Not to mention the huge hit we took on heat rate by blowing all that hot water down the drain...

The condensate cation condos have always been at the high end of the spec, and the specific condos have always been high as well.

Condos have been even higher recently, since we had to keep adding more morpholine to keep the condensate pH from being depressed by the contaminants and massive blowdown.

Oh yeah, lesson #4: Don't let a chemical vendor take the lead in trying to identify the source of a contamination problem. Find it yourself!

 

[bimr]

Also happy to hear that you have solved your problem.

I still fail to understand why you did not notice the condenser leak (with the associated break in high conductivity). Is the quality of the cooling water such that you would not notice the leak?

And why is a relatively new condenser leaking? Shutting down to fix the condenser is not an inexpensive option.

And if your analyses failed to show silica in the demin train, why would you pilot test equipment to take silica out?

Seems like another lesson learned is that some type of structured problem solving should be used. Here is a typical problem solving method for example:

http://www.mapnp.org/library/prsn_prd/prob_slv.htm

Most companies use similar types of programs in their continuous improvement efforts. Maybe you already have one. This type of problem solving may help you solve other problems in your shakedown cruise.

 

[exnavynuke]

Hi bimr,

The chem results were not conclusive. We had samples of raw water, demin water, and condensate sent off to several labs for analyses. Some showed slight hardness, some didn't. Some showed colloidal silica in the raw water, while others didn't.

No tests ever showed *reactive* silica in the demin water, but colloidal silica does not show up on such tests. Colloidal silica is notoriously difficult to detect in low (ppb) concentrations, and only a few labs can provide analyses. Even those lab results have poor resolution. The cutoff for detectability at one lab was 100 ppb colloidal silica. Obviously we could have been adding 99 ppb colloidal silica in our make-up water that the boiler was converting to reactive silica. So the lab results were not the definitive CSI: results we would have liked to have .

Our cooling tower water has an average condo of 2250 umho/cm2, is laden with the usual suspects - ca, mg, na, si, and chloride. The odd thing was that we were seeing *serious* silica contamination, but none (or just hints) of the other circ water constituents... which led everyone to believe it was non-reactive silica getting through the demineralizer, since it the time the problem developed, the demin had no pre-filtration to remove suspended silica.

I think had the tube leak theory and colloidal theory been given equal weighting (disclaimer - which BTW was not my call; I'm just an operator here), the tube leak would have been located much more quickly.

Re: the cause of the tube leak, I anticipate that when we locate the actual leaking tube(s), the cause will turn out to be steam impingement caused by non-diffused/non-shielded drain line penetrations. I will post again when the exact cause is identified.

 

[cub3bead]

No condensers can be such fun! You think you have a tight one and bam you get hit with a tube leak!

Since you operate a combined cycle plant, look in the area where the MS and RH bypasses dump into the condenser. I have seen two Ti tubed condensers fail during startup due to these high energy dumps setting up tube vibrations that caused tube failures. The tubes had also turned a straw to deep purple color which is characteristic of Ti at high temperature in a steam environment. The more purple the hotter it got there.

Are you familiar with a tubesheet cladding process using a 100% solids epoxy? I saw it applied to a condenser to fix another problem and it cut the hotwell conductivity from 10 uS/cm to about 2 uS/cm. Turned out there we a number of leaking tube joints that the cladding covered up. It is a nice system. This type of installation is very common in Europe, where it ws developed. We should see more of it in the US but project developers don't want to spend the $s on something that might be.

 

[exnavynuke]

Hi cub3bead - I believe I've seen the coatings that you are referring to advertised in some of the trade publications. I've no personal experience with them however. They sound sweet though!

So you know, the condensate specific conductivity is still quite high - about 25 umho, although silica has dropped from 70-100 ppb down to around 4 ppb. Cation condo is likewise running higher than it should, at 2-3 umho. So we may have another problem here.

I find it difficult to believe that the small amount of amine we are adding for pH control is responsible for all this conductivity. Your thoughts?

 

[exnavynuke]

Post-outage update. My final post on this issue.

11 main condenser tubes had to be plugged. Several of the plugged tubes had circumferential cracks on the bottom half. The cracks were located midway between tube support plates.

Management asked the O&M tech assigned to the condenser to explain why the failure occurred. He told them to hire an engineering firm to perform an RCFA Good answer!

Since there is no plan to perform an RCFA, we are left to guess what happened. The leading theory here is that the tubes failed due to steam flow-induced vibration (high cycle fatigue), but at this point it's purely educated guesswork.

 

[cub3bead]

I am aware of two units, also 2X1, with down exhaust STGs with the MS and RH bypasses discharging to the hood space above the tubes that experienced similar failures. The condenser tubes are 1" 22 BWG titanium (seawater service). The failures were as you described and in several cases much worse in that 3"-6" of tube were completely missing. The amount of seawater entering the condenser is pretty amazing when you have two 1" tubes free flowing in there.

Our investigation revealed that the EPC had failed to install two rows of solid steel dummy tubes at the top of the bundle as blast deflectors. We ended up packing the top two rows with as much steel as possible and plugging the tubes. The increased mass helps resist the steam flow induced vibration. Notes were put on the drawings to install dummy tubes when retubing occurs.

We also had a bypass temperature control problem evidenced by discoloration of the tubes on top of the bundle, straw color ranging to deep purple. Turns out this is a characteristic of Ti when exposed to high temperatures (300-500 deg F) in a steam environment. In this case the problem was caused by not having enough residence time in the pipe downstream of the desuperheater stations to completely evaporate the spray water before entering the condenser.

Hope this helps.