Dear Ping Pong and Fallah:
Apologies for not providing a more detailed example of one of our several columns.
Here is a sketch of the system:
10-Hour Overfill Column.png 41.53KB
1 downloads
I'll quickly go over your responses:
"I have never heard of a "10-hour liquid overfill response time, it is not mentioned in that wiki article, and I don't see what that wiki article has to do with your case of a tubeburst in a condensor
The accident in Texas did not occur during normal operation, but during start-up when operators were filling up the whole distillation column with hot hydrocarbon liquid without noticing that because level instruments did not work properly.
It is ridiculous to assume that cooling water can run into an operating distillation system while the operators don't notice that for 10 hours."
- The relation of the incident with this problem is that they are both liquid overfill cases. In our project, the client considers that the correct response to the upset may not be done by the operators in much the same way that the Texas operators made several errors in judgement for a long period of time even though they were able to notice the problem shortly.
- The cooling water through a condenser need not to fill up the whole column -- it can fill up the reflux drum and condenser shell for a short period of time (especially if the flow to the column is restricted by control valves)
"If there is a tubeburst in a condensor, the cooling water can only enter the system if its pressure is higher than that of the process system. In that case the reflux drum and condensor will fill with water and no hydrocarbon vapors can condense anymore. The column pressure will go to relief pressure (set pressure + accumulation), which is usually higher than the cooling water pressure, in which case the water level will not rise anymore.
Note also that the 10/13 rule has nothing to do with the problems that you describe."
- Yes. This of course true. The cooling water pressure can reach up to 7.5 barg at the elevation of the condenser (19m). By virtue of 10/13 rule, then I wouldn't have to protect the condenser or the whole system at all from actual overpressure via the cooling water. But, the rate at which the cooling water comes in is very large that it fills up very quickly (even if the path to the column is open).
- Using a pressure vs time and flow rate vs volume analyses, it shows that the cooling water flow through the tube rupture will fill up to the overhead pipe before the hydrocarbon vaporization and pressurization opens the PSV.
- This is where the 10/13 rule comes in PingPong. Even though I don't need to consider the cooling water as being a source of overpressure (since 6 x 1.3 = 7.8 barg > 7.5 barg) it can still put in cooling water in the system because of the difference in operating pressure, elevation, and other hydraulic concerns.
"If in your case the cooling water pressure is higher than the column relief pressure, you can calculate how high the cooling water can rise in the overhead line until there is a pressure balance. A mechanical engineer can then check if the line can carry so much water."
- Yes this is true. The pressure balance can be achieved with the overhead line being filled up. This is why I'm having problems. The main concern is whether or not the line can carry the water. The piping engineer has already stated that it will need additional supports (suggesting that the huge line CANNOT carry water). The additional supports - as I said in my original query, are very expensive and seem impractical for a rare case of tube rupture.
- The query was more about what is the usual industry response for potential cases wherein overfilling can reach huge pipes which are not really designed (support-wise) to carry liquid.
"The impact of a specific failure depends on the specifics of the process design. This kind of failures are considered in the design phase of a project, when developing the P&ID's and specifying relief cases for the PSV's. After that nowadays also a HAZOP is performed on the complete design."
- Certainly, a HAZOP was done. However, during that time a 30 minute response time was considered and no special resolution was required. It was AFTER the HAZOP that the client dropped the bombshell, saying they need 10 hours of CORRECT response time.
- The reason why I'm asking is so I may garner your expert industry experience to be ready to argue with the people in the HAZOP review in case the 10 hours is SO RIDICULOUS that it needs to be stopped.
"If the cooling water pressure would be higher than the process side pressure in overhead condenser, during tube rupture appears the water tends to move toward the reflux drum before running toward column through the overhead line. Then the level will start to build up inside the drum and every safeguard has already been considered to prevent such consequence(s) would come into play for necessary action."
- Unfortunately, if you refer to the sketch I attached above, you will see that the series of control valves can pose as significant pressure drops - restrictions that cause the cooling water to go up preferentially towards the overhead line (becase of the very high flow rate) instead of going down to the reflux and to the larger column volume where it encounters high pressure drops.
- Another unusual thing is that there is a control valve after the condenser. With this at a minimum flow position, the exact situation of preferential flow towards the overhead line is very possible.
I know that this is ridiculous and I know that no one has ever heard of this before -- this is why I tagged it as "unusual". I am building up cases to support the view of "10 hours is ridiculuous". I can prove that it is more practical to have rigorous administrative procedures than a fool-proof design which is overly conservative for an upset event which may not happen within 10 years. But as I am a junior engineer, it might not be perceived as an "expert opinion".
Edited by _luis, 24 May 2014 - 05:43 PM.
Forums
Blogs
Articles
File Library
FAQ
Search
FB
