11 replies to this topic

[LVBob]

Greetings!

I’m currently a Process Engineer handling gasoil hydrotreaters at our refinery. I've only been a Process Engineer for about 8 months. I know that you need as much details about the problem and I hope this is informative enough. If you need more, I'd be very willing to give more details.

Here's some background about the problem:

We noticed a steady increase in the pressure drop between E-1(HHPS Flash Gas-Treat Gas H/E) to E-2 (Airfin cooler) beginning on August 14, 2011, reaching values as high as 600 kPa (versus a typical operating pressure drop fluctuating between 120-180 kPa). The pressure taps are placed at the inlet of E-1 and outlet of E-2. We first injected water at the inlet of E-2 but saw only a minor drop in dP. We then diverted the water to the inlet of E-1 and noticed the drop more significantly. Only a minute or two of water washing is sufficient to bring the dP down to 90 kPa. After water washing, the dP would continuously increase until we decide to water wash at 350 kPa. It was only after the third wash when the dP stabilized at around 130 kPa. We concluded that it was fouling at the tube-side of E-1. Our initial evaluations using sublimation curves show that the most probable cause of fouling is the presence of chlorides in the vapor. We suspect that this may have reacted with ammonia to form ammonium chlorides in the exchanger’s tube-side. I attached a PFD of the process for your reference.


What we need to find out is where the chlorides may have come from.

Here’s what we did:

• At first we thought that the reformer makeup gas (hydrogen-rich) was having a chlorides breakthrough. Draeger tube tests showed chlorides of about 0.2 to 0.4 ppmv (versus typical 0.0 ppmv). On September 25, 2011, the adsorbent for the makeup gas chlorides treater was replaced with fresh stock but the exchanger was still experiencing a steady increase in pressure drop until our third water wash on around October 19, 2011. Since that third water wash, the dP has stabilized to around 130-200 kPa. Furthermore, the adsorbent’s supplier explained that there was never really any chlorides breakthrough. The readings were due to interference from contaminants. They proved this by pushing makeup gas through a small bed of NaCl and took measurements of the filtered gas using Draeger tubes. These showed 0.0 ppmv.
• We also checked the feed for any chlorides. We sent samples of straight-run feed, light cycle oil (LCO), and light vacuum gasoil (LVGO) to a third party lab to test the samples for total chlorides via UOP 779. The results showed <1 ppm of chlorides in each of the samples (beyond the detection limit of the test).
• We checked if there were any idle lines that could have been passing-thru and may have been injecting raw water, or any other substance with chlorides. There were none.
• We checked our shipping records for batches of crude and plotted the duration of their consumption against the beginning and end of the fouling incident. None of them matched. Furthermore, the chlorides content in the sour water at our pipestills’ overhead accumulator showed typical chlorides content. In other words, there was nothing unusual at the pipestills.
• We checked our slopping from the other units and found that benzene slop from our BTX unit (which uses an ED Sulfolane process) was fed into the pipestills on August 10, 2011. Feeding of the slop, however, ended at around August 22, 2011. This did not coincide with the whole duration of the fouling incident.

We were wondering if you had encountered any similar incidents in other refineries. We would also like to ask the following questions:

• Were we correct to conclude that the rapid increase in dP across the heat exchangers was due to fouling by NH4Cl? We already eliminated NH4SH as a foulant and there were no unusual process parameters during the time. Is there anything else we should look into aside from fouling that may have caused the rapid increase?
• Where could the chlorides have possibly come from? We have already eliminated the hydrogen makeup gas as a source of chlorides based on the adsorbent supplier's "proof" and the timeline of events.
• If the chlorides came from the feed, what mechanisms could have been present that could cause the chlorides to form HCl? How spontaneous are these reactions? At what temperature and pressure will the reaction occur at a significant rate?
• Do you know of any reference material/articles/consultants that can discuss this in length?

Hoping for your kind assistance.

Thank you very much.

Attached Files

•  PFD.png   25.84KB   24 downloads

Edited by lvroperos, 27 January 2012 - 03:49 AM.

[Himanshu Sharma]

Can you explain "We already eliminated NH4SH as a foulanIt"

Have you checked Nitrogen content in the feed as compared with Nitrogen content of Design feed.?

I envisage that fouling is observed due to Ammonimum Bisulfate ,this often is a problem if your HP air cooler or connected piping is not designed to withstand the current concentration of NH4HS(higher than salt conc for which unit is designed).

Employ the wash water quantity as highest of the of 8wt% ammonium bisulfate max(if HP REAC MOC is duplex2205), 5%FF wash water to feed ratio min or 20% water in liquid phase at REAC inlet.

I hope that this shall serve the purpose.

[LVBob]

Hello Himanshu.

Thanks for the quick reply.

What I meant by "eliminated NH4SH as a foulant" was that we concluded that NH4SH could not have been causing the majority of the deposition at E-1. We made calculations of k = (pNH3)(pH2S) and plotted this with temperature in a sublimation curve. We found that the point lied outside of the deposition region (please see attached). It was just too hot at E-1 for any bisulfides to deposit. As for NH4Cl, however, the conditions were just right. Deposition was possible (please see attached).

Just to clarify, we injected water into the inlet of E-2 (air cooler) first. The dP didn't drop signficantly. We then injected water at the inlet of E-1 (S&T H/E). This caused the dP to drop to 90 kPa (from 500 kPa) within just a few minutes. Also, we placed a local pressure gauge between E-1 and E-2. We confirmed that the fouling was at E-1 because much of the dP was there.

We didn't check the nitrogen content. Our lab doesn't have the capability to do so. But as additional info, the amount of NH3 in the vapor (per material balance) is 0.4388 kmol/hr or 0.032 mol%. We use Cobalt-Molybdate (CoMo) catalyst and should be selective for desulfurization.

By the way, how do you know if an exchanger is not designed for this level of fouling? What do you look at? What calculations should I be doing? Should calculating for U or fouling factors be sufficient?

Thanks for the help.


Appendix:

Here are some details:

E-1
Tube-side (HHPS vapor) Inlet temperature = 230 - 250 °C (normal operating)
Tube-side (HHPS vapor) Outlet temperature = 161.3 °C (design SOR)
Inlet pressure = 4350 - 4380 kPag (normal operating)

E-2
Air cooler, pitch-control
Inlet temperature = 161.3 °C (design SOR)
Outlet temperature = 28 - 35 °C (normal operating, temperature-controlled)
Outlet pressure = 4200 kPag (normaal operating, pressure-controlled)

Attached Files

•  nh4sh.PNG   20.53KB   36 downloads
•  nh4cl.png   143.3KB   30 downloads

Edited by LVBob, 28 January 2012 - 03:16 AM.

[Himanshu Sharma]

Dear LVBob

Original bundle construction in these exchanges has commonly been carbon steel or Cr-Mo steel, if dictated by high temperature hydrogen attack considerations. Elevated corrosion rates have often necessitated an upgrade to alloy.

as a general thumb rule

less than 3 wt% ammonium bisulfate go for KCS
between 3~8wt% calls for duplex 2205
more than 8 wt% moc should be incolloy
This decision is a tradeoff between wash water rates and MOC (opex vs capex).

you can measure this salt concentration in sour water from cold separator.

as you have actual ammonia concentration available with you ,please check it with design number in stream summary for original design and compare .

Additionally ,For heat exchager and the aircooler velocity of reactor effluent becomes an important parameter for design.

My concern here's more than pressure drop is ex changer fouling ,have you opened unit for inspection off late ? may be you can expect a greater problem of corrosion there !!!

During normal unit operations, chlorides can be introduced into the reactor section via either the feed or makeup hydrogen streams. The chloride input can result in ammonium chloride deposition and corrosion problems in both the reactor and fractionation sections.

Experience indicates that chloride attack will occur if NH4Cl deposits coexist with the vapor phase at or near water dewpoint conditions. Corrosion susceptibility develops when poor oil/water separation provides a mechanism to transfer H20, NH3, H2S and CI from the reactor section to the fractionation section. As water is vaporized in the fractionator feed preheat circuit, ammonium salts precipitate. Corrosion/cracking potential becomes problematic at those locations where the ammonium salts exist in concert with water dewpoint conditions.

Edited by Himanshu Sharma, 28 January 2012 - 03:54 PM.

[LVBob]

Dear Himanshu.

Thank you for the very comprehensive reply. I'll get back to you as soon as I check out your recommendations. Your advice is very much appreciated.

[LVBob]

Hello again.

I recalculated the partial pressures for each of NH3, H2S, and HCl, and compared them with the design values (please see attached spreadsheet). What's fascinating about this is that the designers seem to have not considered HCl in the design. The species isn't present in the design's material balance. Nevertheless, ammonia partial pressure during that time was below the design EOR value. We can safely conclude that there wasn't anything unusual with the nitrogen in the feed.

Also, the reactor is operating at 310 to 330 °C and about 4830 kPag (inlet)/4620 kPag (outlet). The temperatures are beyond the range of the sublimation curves that we have at the moment. Is it possible for NH4Cl to deposit under such conditions, assuming that the partial pressures that I calculated are approximately the same at the reactor?

Lastly, the pressure drop across the reactor bed was normal during the time. There was definitely no fouling at the reactor.

I'm requesting maintenance to open up the unit in a few months. I'll update you on what we find.

Thanks again for the help.

Attached Files

•  Partial Pressure calculations.xlsx   10.51KB   71 downloads

[Himanshu Sharma]

Dear LVBob

Kudos for working so hard and i am astonished for you to gather this kind of wisdom in first 8 months of process engineer role.

Getting to business,HCl is normally not reported in the stream summary you get from process licensor but is generally considered for wash water calculation for salt removal.

can you check that whether chlorine+chloride content is listed in the feed specification, and compare the design values with actual conditions.

Talking about simulation curve ,theoretically they do predict a great deal of information but what happens inside a reactor is 'predictions' there could be a local coldspot or any local catalylization or nucleation of salt crystals that can lead to salt deposition ! anything in this world and we just keep guessing 'may be & could be's'.

[LVBob]

Hello Himanshu

Thanks for the compliment. I don't get much of that these days. Haha.

Alright, here's what I found. There was no mention of any chlorine or chloride species in the feed specifications nor in the specifications for the makeup gas.

By the way, I'm posting a revision of my calculations. I carelessly switched Kh at std conditions for -delHsoln/R. It's only effect is to reduce the partial pressures. Nevertheless, based on the curves, NH4Cl is still the likely foulant.

Attached Files

•  Partial Pressure calculations.xlsx   10.49KB   47 downloads

[LVBob]

Hello Himanshu

I checked the temperature profile of the reactor during the fouling period and I'm relieved I didn't find any local cold spots. The temperatures varied by about ± 0.5 to 1 °C among temperatures in each layer of the reactor. Nothing unusual there.

I also checked the shift reports between August and October (the whole duration of the fouling). I found out that our LVN Isomerization Unit (LIU for short; also receives reformer makeup gas) also had a chlorides breakthrough only 6 hours prior to the start of our fouling incident. Polychloroethylene injection rates were normal during the time but they found chlorides in the range of 10 to 50 ppm at the deisohexanizer (DIH) tower receiver liquid (iC5/nC5).

They suspected that chloride-rich gas from the stabilizer tower somehow entered the DIH tower because the stab tower was operating at a lower level than usual. To minimize ingress of chlorides and remove them from the DIH, they increased reboiling at the stab tower and vented the DIH receiver to the flare.

What I find suspicious about all this is that the fouling ended right about the same time that the LIU was shut down on October 17, 2011 because we had steam supply issues then.

The LIU feeds fuel gas to the header, which we use to pressurize the hydrotreater's feed surge drum.

Could HCl (from the fuel gas) have dissolved into the feed at the surge drum? Do you know of any VLE data for HCl and diesel? This must be why we never detected any chlorides form the individual feed streams. We never tested the mixed feed at the feed surge drum.

Thanks for the help.

[Himanshu Sharma]

Hi LvBob

was out on a business assignment and could read your reply just now.Somehow the events that you are describing feels connected isn't that watson .If we need to figure out "How" we have to work hard a lil bit.

Normally the ISOM off gases are sent to LPG for recovery in CCR recontacting section and reformer H2 rich gases are supplied to a common compressor in ISOM to supply the same to NHT as well.

Can you email me the PFD/P&IDs for integration of NHT,ISOM & CCR or share it here if there's no confidentiality issues.

[LVBob]

Hello Himanshu

I'm sorry but I can't share the PFD/P&IDs. Company rules you see. But suffice to say that the off gases from ISOM go to the fuel gas header. The LPG units take in gas from the FCC and I think that the LPG units' capacity can't take in any more off gas from ISOM. As for the H2 makeup gas from CCR, it's supplied via a common compressor at the CCR. There are other makeup gas compressors at the gasoil hydrotreater as well - three in parallel with one as spare.

[LVBob]

Just a little update:

The chlorides didn't come from the LIU. The breakthrough was only in the product. The fuel gas is scrubbed with caustic before routing it to the fuel gas header. They measured the chlorides content during that time and they didn't find any.

We recommended closing the case because of the extensive amount of resources (in terms of man-hours) that this is costing us. Nevertheless, we've developed an action plan should this recur.

*sigh of relief* This is finally over. I could start with my other projects.