Sep-2021

Increase reliability and profitability in delayed coking units

Global refiners are always seeking new techniques to optimise their refinery assets with the purpose of maximising their profitability.

David William and Sabrina Feldmann
Berthold Technologies

Article Summary

Most refiners look for methods to convert lower value residual into higher value products. In other words, refiners are searching for bottom of the barrel conversion technologies such as visbreaking, delayed coking and resid hydrocracking, either ebullated bed or slurry hydrocrackers (Sawarkar et al, 2007). Since delayed coking is one of the most profitable units in the refinery and it has a lower initial cost than some of the other heavy oil conversion technologies, many refiners are turning to this proven conversion technology (Jaguste, 2016). The coking process dates back as far as the later 1920s, around 1929 when Standard Oil in Whiting, IL started the first delayed coker unit.

Delayed coking overview
The heart of the delayed coking unit is the drums where the actual cracking of the hydrocarbons takes place. The unit is the only semi-continuous batch process in the refinery, which means that feed is continuously switched between two, or sometimes more than two, drums in a time-based cycle, typically 12 hours (Jaguste, 2016).

In the simplified view, fresh feed, usually from the vacuum tower bottoms, is fed to the bottom of the coker fractionation tower (Figure 1, item 3) to help pre-heat the feed and is mixed with the bottom hydrocarbons (product that was not fractionated out) of the fractionator tower. This mixed resid is then fed to the furnace (Figure 1, item 1), where the resid has enough energy applied to thermally crack the large hydrocarbon chains. Since the required time for this cracking is 30−45 minutes (hence the name delayed coking unit), the resid is fed into the coke drums (Figure 1, item 2) where it remains until it is thermally cracked (Sawarkar et al, 2007). After the resid is cracked the resultant hydrocarbons are vaporised and leave the drums to be cooled and fractionated into different side streams in the coker fractionator (Figure 1, item 3). Some of the vapours are not fractionated into the side streams and therefore mix with fresh incoming feed to be recycled back to the furnace where they are further cracked. This is referred to as recycle ratio — the recycle feed compared to the fresh feed (Motaghi et al, 2010).

At the end of the approx. 12h cycle the coke drum is filled with a solid residuum of coke. To remove this coke, a high-pressure water drilling system is used to cut the coke block into chunks, that can be then removed.

Level measurements to control the process
One of the most critical measurements in the delayed coking unit is the level measurement inside the coking drums. Due to the conditions inside of the drums this measurement is a challenging application. As the heavy hydrocarbons are thermally cracked, the resulting hydrocarbons are then converted into a vapour as a result of the high temperature. As the vapours escape the viscous liquid, they tend to create a foam layer. This foam layer can vary depending on several parameters such as the operating temperature, pressure of the drum, type of crude or charging rate. To increase throughput of the unit, one of the most important objectives is to fill the drum higher safely and reliably (Sawarkar et al, 2007). For this reason, it is necessary to reliably measure the foam front in the drum to ensure that it does not go over into the overhead lines and/or fractionator tower, in other words, a so-called foam- over should be avoided. A foam-over in a delayed coker unit is a costly event, not only due to the loss of production but also the manual labour required to clean the coke out of the overhead lines and fractionation tower. This cleaning can take up to two or three weeks depending on the severity of the foam-over. In the meantime, many different types of technologies have been tried and tested to measure this level. Among them are differential pressure (DP), radar and nuclear. Due to the extreme process conditions inside the drums, DP and radar units tend to either fail or turn out to be extremely unreliable. Thus, almost all coke drum level measurements are some type of nuclear measurement devices.

Different types of measurement devices
The first type of nuclear gauge that was tried was a gamma switch. This consisted of an internal source and an externally mounted detector (Figure 2). The internal source was required because of the large diameter of the drum and the substantial wall thickness. Furthermore the detector technology at the time was not sensitive enough to measure the lower radiation fields as is possible today.

This type of measurement did not last long due to the complications posed by the internal source well and drilling of the drum. Customers did not want to take the risk of losing a source during cutting of the drum. The shift in the industry was initially towards a neutron backscatter measurement (NBS) since both the source and detector are located outside the drum in the same housing. Consequently the mounting of the device was much easier.

The operation of the NBS also has some limitations due to the principle of measurement. While the NBS is used to measure level in the delayed coke drums, in reality it measures the concentration of hydrogen of the material inside the drum. As water has more hydrogen than resid, resid has more hydrogen than the foam, and the foam has more hydrogen than the hydrocarbon vapour, the device is used as an indirect measurement of the level. The NBS emits neutron radiation in all directions because it is very difficult to collimate neutron radiation.

To measure the level inside the coke drum with the NBS, two things must be given (see Figure 3):
1. The neutrons must be scattered back to the detector, since the source and detector are in the same housing.
2. Secondly the neutrons must be thermalised or slowed down by the process material. If the neutrons are not thermalised, they have too much energy to be detected by the sensor and travel through the device undetected (Hart, 2014).

One of the most effective means of thermalising neutrons is using hydrogen as a moderator. Hydrogen is very effective in thermalising or slowing neutrons down since they are physically the same size. To illustrate: consider billiard balls, if the cue ball hits another ball, half the energy of the cue ball is transferred to the other ball (comparable to a neutron colliding with hydrogen nuclei). If we now assume that a cue ball hits a bowling ball, the cue ball bounces off basically at the same speed, as very little energy is transferred to the bowling ball. This is comparable to the collision of a neutron with a carbon nucleus. The distance a neutron can travel after being thermalised is very limited, therefore the NBS measures approximately 450mm into the drums. Thus, the area a NBS can measure is a sphere with a radius of 450mm. As the hydrogen content increases in front of the detector, more neutrons are thermalised and scattered back to the detector. As already mentioned before, the hydrogen content of the different phases inside the coke drum are different. So, in principle NBS measures the amount of hydrogen in front of the device. By monitoring the output changes in the detector, an operator can ascertain the level inside the coke drum. This relationship between the amount of radiation sensed and the level is a direct proportionate relationship — less radiation detected equals less hydrogen or lower level, more radiation detected equals more hydrogen or higher level

Typical coke drum arrangements
A typical coke drum will have either three or four NBS located at different elevations on the drum (Figure 4). These elevations are normally 3m, 4.5m and 7.5m down from the top tangent. The 7.5m is used to start anti-foam injection, 4.5m is typically the maximum coke bed density and the 3m is basically a high-level switch. Operators would use the NBS like switches on the side of the drums to inform them if the level has reached a certain elevation. The output of the NBS is supposed to inform the operator if the material at the elevation is hydrocarbon vapour, low density foam, high density foam, coke, or water. Theoretically this is possible since the hydrogen content is different in each of these materials. The main issue with this assumption is that the NBS measures over a range of approximately 800−900mm (400−450mm up and down) and the NBS output assumes that the range is covered by a single material, which is not necessarily the case. For instance, if in one case the range was covered by 20% coke and the rest was hydrocarbon vapour (no foam), you might get a reading of approximately 40%. At the same time, if the range of the NBS is covered with 70% high density foam and 30% low density foam, you might also get a reading of approximately 40%. This is due to the total amount of hydrogen measured over the entire range.