M.M. Naklie
Mobil Exploration & Producing Services Inc.
Dallas
L. Pless
Tru-Tec Inc.
Houston
T.P. Gurning, M. Ilyasak
P.T. Arun Natural Gas Liquefaction Co.
Sumatra
Radiation scanning has been used effectively to troubleshoot the treating towers of the Arun LNG plant in Sumatra, Indonesia. The plant is one of the world's largest such facilities.
The analysis was part of an investigation aimed at increasing the capacity of the treater section of the plant.
SIX TRAINS
The Arun LNG plant, built and owned by Pertamina, contains six treating and LNG trains. The plant is operated by P.T. Arun Natural Gas Liquefaction Co., a nonprofit Indonesian corporation owned 55% by Pertamina, 30% by Mobil LNG Indonesia Inc., and 15% by Jilco (Japan Indonesia LNG Co.).
Each treating train consists of a Benfield unit, which includes a hot potassium carbonate system followed by an amine system (DEA). The carbonate system removes the H2S in the feed gas and reduces the CO2 content from 15% to 4,000 ppm. The amine system reduces the CO2 content to 50 ppm.
At high rates it was noted that some of the trains performed better than others.
Train 6 recently had a carbonate absorber liquid redistributor and a carbonate regenerator top distributor upgraded to new designs, and the performances of both were less than expected. In order to analyze better what was happening inside the towers during operation, radiation scanning was used to troubleshoot the absorbers and regenerators of the Benfield and amine systems.
Radiation scanning is a tool which, in addition to tower differential pressure and product purity, can aid in diagnosing tower performance.
Radiation scanning, also termed "distillation column scanning" or "gamma scanning," is the procedure of studying the process material inside a process vessel by moving a radioactive source simultaneously with a radiation detector along the exterior of the process vessel (Fig. 1).
This technique is based on the principle that an increase in material density will reduce the radiation signal, or a decrease in material density will increase the radiation signal. The result is a density profile of the inside of the process vessel.
For distillation or absorption columns, correlating the changes in density seen in the scan profile with column internals and column elevation yields pertinent information on the column's performance and physical condition.
THICK-WALL VESSELS
Tru-Tec Inc., Houston, one of the few companies that perform radiation scanning, was called in for the Arun scanning primarily because of its experience with packed columns, which are more difficult to interpret than trayed towers. The carbonate and amine absorbers and regenerators are all packed columns, with diameters ranging from 10 to 18 ft.
In addition,Tru-Tec has experience scanning through thick-walled vessels, up to 4 in. thick.
The carbonate absorber wall thickness is 3 in. This wall thickness required a stronger radioactive source, Cobalt 66, than is normally used to give a clear representation.
It should be noted that, although a radioactive source is used, the radiation exposure to plant personnel is negligible. Standard safety procedures, including lead shielding and posting restricted areas, are practiced.
In addition, radiation column scanning poses much less exposure risk than typical weld X-raying. Because of the wall thickness, a 300-millicurie cobalt source was used for the Arun LNG scanning instead of a typical scanning source of 50 millicuries.
The normal radiographers' source is 55 curies, a thousand times that typically used for scanning.
Four scanlines were made of each packed column at the Arun LNG plant, two scanlines in one direction and two more scanlines at 90. Each complete column scan took the two-man crew 4-6 hr to complete, depending on the height of the column.
For packed columns, as with trayed columns, scanning provides an on-line inspection for column internals. Scanning can show if the column internals are indeed in their proper place. Problems such as damaged or missing distributor trays, collapsed beds, crushed beds, etc. have been diagnosed successfully with scanning.
Another important evaluation of a packed column from scanning is the degree of vapor/liquid maldistribution through the packing.
With ideal distribution all four of the scan data plots through the packing will coincide, or overlay on top of each other. Coincidence of data indicates a uniform density through the beds and across distributors and collectors in the column.
Problems are diagnosed when nonuniformity is observed from the scan plots. Any spread among the four plot curves is an indication of maldistribution, with liquid bias associated with lower radiation intensity, and vapor bias associated with the higher radiation intensity.
FEED-GAS DISTRIBUTION
Fig. 2 is a plot of the data (counts of radiation/6 sec vs. column height) from the scan of the Train 6 carbonate absorber. The scale through the packed beds has been compressed; normally, a plot for a column this large would be on multiple, continuous pages. (Curve lines in Figs. 2-8 represent consecutive scans.)
Fig. 3 is a plot of the lower half of the Train 6 absorber. The scans indicated possible damage to the feed gas piping at the bottom of the absorber.
These plots are compared to Fig. 4, the plot of the lower half of Train 3. Train 3 had normal scan results for the gas distributor.
The piping on Train 6 was inspected shortly thereafter during a turnaround. Loose flanges were found on the feed-gas piping allowing leaking from the piping distributor.
The rates were increased to maximum on the treaters, and scans were made of the Train 3 and Train 4 absorbers at the same rates. The scan results for the bottom half of Train 3 and Train 4 carbonate absorbers are shown in Figs. 4 and 5, respectively.
The tower internals are identical on these two columns, except that the Train 4 absorber has larger-sized packing in the bottom bed.
The scans of the bottom beds indicate that the bottom bed of Train 3 is nearing its capacity limit at maximum rates. Fig. 4 shows where liquid stacks up from the top of the bottom bed through the redistribution area.
Fig. 5 shows Train 4 did not have this problem. Note the clear vapor space between the redistributor pan and the top of the packing in the Train 4 absorber and the lack thereof in the Train 3 absorber.
The bottleneck of liquid in the first few feet of a packed bed, like in the Train 3 absorber, normally indicates that the packing cannot handle the process load.
The scans of both carbonate absorbers indicated foaming in the lower part of the bottom bed of packing. Vapor spaces below the bottom bed were denser than expected. This was further evidence of foaming, whereas flooding would have had a sharper increase in radiation intensity.
TOP LIQUID DISTRIBUTOR
Figs. 6 and 7 are the scan results for the top half of the Train 3 and Train 4 absorbers, respectively.
The scans showed that the Train 4 absorber had good vapor/liquid distribution in its top bed of packing. By contrast, the scans showed that the Train 3 absorber had poorer vapor/liquid distribution through its top bed of packing.
Fig. 8 is a graphical representation of the scan data from the top bed of packing for both absorbers. The graph is a statistical treatment of the scan data which aids in viewing how well the data coincide and aids in identifying the pattern of flow maldistribution.
Fig. 8 confirms what can be seen by closely comparing Figs. 6 and 7: that the process density through the top bed in the Train 4 absorber was fairly uniform and that the process distribution was better in the top bed of the Train 4 absorber than in Train 3.
Fig. 6 shows that the top distributor area in the Train 3 absorber had liquid stacked up, and that Tray 1 was holding a high liquid load. Because this area of the absorber has fouled historically with carbonate buildup, it appears that the Train 3 absorber had a fouled distributor.
REGENERATOR DISTRIBUTOR
The carbonate regenerators also were scanned. The scans showed poor distribution in all the regenerators. The top flashing feed distributor in Train 6 is a different design from all the other trains, and it was anticipated to give better distribution.
Although the scans showed slightly better performance for Train 6, it was still rather poor. The scans identified a liquid distribution problem but the cause was not readily obvious. Inspection of this distributor during the turnaround revealed mechanical damage.
The scanning was ineffective in differentiating the performance of the absorber liquid redistributor in the various trains. Apparently the close distances between the hardware zones (4-5 in.) and the complexity of the redistributor hardware resulted in similar scans, so were inconclusive.
A technique to handle this type of problem may be to scan for comparison when the column is empty or not operating.
Copyright 1990 Oil & Gas Journal. All Rights Reserved.