5 replies to this topic

[Mateo_1974]

Hola, I have recently joined a new gas plant which exports to our domestic gas network. At our site we have a molecular sieve for dew point control with a tight spec of -100°C as we have a cryogenic nitrogen rejection unit due to the high nitrogen in our feed stock.

We have three molecular sieve beds (4A) and have had no real operational issues with these beds, however we have a site practice to flare gas to maintain a minimum flow through the beds when we are partially shutdown. The concern is that flow below the minimum flow will cause channeling and therefore premature breakthrough.

I do not have this concern, having worked in a previous plant that we never focused on channeling in adsorption and only during regeneration.

We operate two beds in adsorption and one in regeneration/bypassed.

The normal adsorption flow per bed is 8 mmscf/hr and a regeneration flow of 1 mmscf/hr with a cycle time of 24 hours.

Our vendor quoted a figure of flow below 1.5 mmscf/hr would cause channeling, and therefore if we do not have forward flow we would have to flare 0.5 mmscf/hr (we can flow off flow to one bed). I don’t see the risk and would assume operating at 1 mmscf/hr  (13% of the normal flow) would still be flowing through a reasonable portion of the bed (surely more than 13% of the bed), I can understand that we would not be able to use the entire capacity (we use a worst case assumed uptake capacity on our bed of 11 wt%, though generally we operate at about 8 wt% before it goes into regeneration).

I have been unable to find any literature on channeling in adsorption and most of it focuses on during regeneration (an specifically hot regeneration) to ensure all the moisture is removed right to the edge of the bed.

Does anyone have advice on how to demonstrate that the risk of breakthrough is low due to channeling in adsorption? Or a calculation to determine the portion of bed a channel would pass through?

Thanks,

Mateo

 

[Mateo_1974]

Please not that the figure for 1.5 mmscf/hr causing channeling is in the adsorption flow direction. For hot regeneration our low flow alarm/channeling limit is 0.85 mmscf/hr. We have flowed this low before and had no issue heating or cooling the bed. 

[Pilesar]

Packed beds have more open space at the wall and there is the lowest pressure drop. With low flow, the wall adsorbent will see more flow in proportion to the center. Search 'wall channeling in adsorption' or similar term. Example scholarly paper is here: https://www.aidic.it/cet/16/52/074.pdf

That is the theory. In practice, there is no reason I see that the overall bed run will be shortened. If the regen is based on 'time in service' then I tend to side with you that minimum flow restriction may not be needed. If you can monitor outlet dewpoint and have a fresh bed ready, why not run some experiments?

[Art Montemayor]

Mateo_1974:

 

What make and model of dew point analyzer are you using to measure the -100 ^oC atmospheric dew point spec value?

 

I never was able to measure a dew point that low and I'm interested in seeing how the instrument industry has improved.

Also, how does your adsorbent vendor measure the amount of gas feed channeling within your beds?  Is this based on a field measurement or is this based on pilot plant results - or theory?   Your comments on this would be helpful.

 

Thanks.

[Mateo_1974]

Thanks Pilesar, I found a another CFD article that also explained some of the wall channeling concept (in this simulation the failure of packing rope caused severe wall channeling) - unfortunately it did not cover low flow channeling. Interesting in the article you shared, it shows some paths operating at 8 times the inlet velocity (which was a bit over three times the average velocity), in my case on turndown we are operating at about 1/8th of the normal inlet velocity - so in my case these accelerated paths would only be back to the normal velocity. So it does give me some comfort. 

It will be interesting to see if anyone has some of their own practice they can share. Unfortunately running experiments at our site is not easy - this would require us to run at low production for a day and the production cost would out weigh the benefit that this experiment could demonstrate. A frequent time we run at these low rates is following a turnaround and due to our moisture analyzer configuration (long pipe run), it takes about a day to dry out completely so we are running blind.

 

Art, we use an Ametek "quartz crystal microbalance" - I think it is a 3050 series. Their white paper shows readings down to 0.01 ppmv. In practice not sure, though we get concerned if we saw ours trending up, especially if there is a trend above 0.05ppmv rather than data bounces around. Though we have never had breakthrough to see this in practice. 

 

Our vendor has there software, I believe a bit of a black box. No doubt it would be based on theory and pilot plants, 

[Art Montemayor]

Mateo_1974:

 

We don’t know what gas or gas mixture you are feeding your adsorption unit and what you are removing – or co-adsorbing.  I presume you are simply removing water content from a high nitrogen content gas and are doing this prior to feeding a cryogenic separation tower that separates out the nitrogen by fractionation.  We also don’t know the physical design of your adsorbers: are the 3 vessels regenerated with a dry regen gas at a minimum of 500 ºF?  Is the moist feed gas fed in a downward flow and the regen gas in an upward flow?  Are the regenerated beds cooled with the same dry regen gas?

 

How are the Mol Sieve beds supported?  Do you use support grids, like the attached Johnson literature?  Or do you use ceramic balls?  I assume each vessel has one randomly packed bed of Mol Sieve and no separated sections.  What are the Mol Sieve particles?  Spheres, pellets, granules?  What is the measured pressure drop across each bed at startup, mid-running time, and at the end of each drying cycle?

 

I agree that any possible channeling during regeneration might be more harmful because it would lead to possible subsequent break-through when the regenerated beds would be put into adsorption.  However, it might be more difficult to monitor pressure drop during regen due the relative lower flow rates and superficial velocity.

 

Basically, there are a lot of parameters that can affect a deficient adsorption cycle and ultimate break through of moisture in the product gas stream.  And to make matters worse, there is no absolute or special way to identify, measure, and much less monitor any channeling taking place.  In all my years of operating, designing, and building adsorption units in the field I was only able to use logic and what I considered to be the best practices in the art.  During the 1960s and 70s we only had basic information on applying Mol Sieves.  In converting some of the compressed gas plants under my charge I used both Mol Sieves and Activated Alumina.  I used 20 ft/min as my design superficial adsorption velocity and a minimum of 500 ºF for regen gas when drying for cryogenic downstream operations.  I always used cryogenic waste gas for regen.  With this basic design I easily achieved -90 to -100 ºF dew points – the max at that time that could be measured.  I always used metal grids to support my beds and tried my best to measure the packed bed central zone(s) for temperature readings.  I used the Ergun equation to figure my bed pressure drops and always used hold drown devices and downflows to minimize bed lift and movement.  In all the dryers I operated and built I never obtained any dusting, break throughs, or suspicions of channeling.  I never dropped or changed out any adsorbent before 5 years of service on a 7 day/24-hr operation routine.

 

I don’t understand your explanation of how you operate your adsorbers so it’s difficult to comment on how I would find out if there is any issue to worry about concerning channeling.  In my experience I would not give the potential for channeling around the beds any importance unless I had the actual process calculations and the detailed fabrication drawings of the designers and fabricators.  Without knowing their parameters and what they designed for, I could not arrive on what to expect as limits or potential issues with your system and its operation(s).  With today’s improved and sophisticated field instrumentation it is much simpler and easier to analyze and trouble shoot a process operation.  Basically, if you are capable of measuring and monitoring the regen temperatures within the central core of your adsorption beds in the bottom, central, and top sections you should be all right in being able to identify that you are OK with your beds being in good, regenerated shape and safe from any subsequent break through.  Locating thermometers at suspected channeling sites and confirming their readings compare with core readings will prove if channeling occurs.  I would certainly trust and abide by that kind of empirical field data rather than relying on a computer program based on theory, unknown experimental data or algorithms.  Until someone invents a trustworthy channeling flowmeter, I wouldn’t rely on anything else.  To the degree that you can safely and thoroughly regenerate all of your adsorbent beds, you will always be able to expect designed performance from them.  And the most basic and trustworthy information to ensure that, is from your strategically located regen temperature indicators.

 

I sincerely hope my experience and comments may give you some ideas or helpful confidence in your operations and that you are able to overcome this issue.

 

 Johnson Support Grids 01.pdf   421.08KB   77 downloads