120%
10% 12% 14% 16%
Liquid from VO Fossil liquid
C from VO H S + COx + H 0
100%
80%
60%
0% 6% 4% 2% 8%
40%
20%
0%
0% 20%
40%
60%
80% 100% 120%
0% 20%
40%
60%
80% 100% 120%
Veg e table oil in feed , vol%
Veg e table oil in feed , vol%
Figure 5 Product yields (C 3 from VO and H 2 S + COx + H 2 O)
Figure 4 Product yields (liquid from VO and fossil liquid)
those originating from vegetable oil, as illustrated in Figures 4 and 5 . Figure 6 indicates that the predicted percentage of HDO calculated with the model closely matched the exper- imental results. By calibrating the model, commercial performance can be predicted, which has been done for the case with the 20% soybean oil. Case study with 20% soybean oil Based on the properties shown in Table 1 , 100% straight- run gas oil (SRGO) is considered the base case. Given the quality of the SRGO and the test conditions selected, we anticipate a cycle length of at least four years can be obtained. For predicting the performance of 20% soybean oil, the catalyst activities from the test were taken and a catalyst deactivation rate, assuming a four-year cycle length for the
base case. Production of ULSD with 8 ppm S was simu- lated. For the 20% soybean oil case, we estimated the start-of-run (SOR) weighted average bed temperature (SOR WABT), exotherm and end-of-run (EOR) conditions. Additionally, we calculated the percentage HDO, which was found to be higher than in the test, as shown in Figure 6 . This is related to the fact that commercial units are adia- batic and operate at a lower reactor inlet temperature, the temperature at which the soybean oil is converted. The estimated cycle length for the 20% soybean oil test conditions appeared longer than the base case. This is because the 20% soybean oil blended feed sulphur con- tent is lower, and the CO formed creates a low CO partial pressure because of the increased H₂/oil ratio. However, we expect the CO partial pressure in commercial units to be much higher because a significant part of the CO formed will be recycled. A longer cycle length is only achievable when all the treat gas going to the hydrotreater is fresh make-up gas without CO. In almost all cases, the treat gas consists, for a major part, of recycle gas containing some CO. The CO in treat gas needs to be controlled by purging some of the recycle gas. Figure 6 shows that a higher purge gas rate results in a lower CO in treat gas, reducing the required operating temperature and increasing cycle length. Still, the hydrogen loss comes at a cost. Consequently, the question must be raised: “What is the most optimal purge gas rate?”.
Different feedstock properties
Feed type
SRGO 0.8569 14,848
Soybean oil
20% blend
Density @ 15°C
0.9255
0.8716 11,003
S, ppm N, ppm
267
213
SimDist, °C 5%
221 403
553 614
229 606
95%
Table 1 Comparing performance of different feedstocks
7%
35
0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100%
30
6%
Test result Commercial unit Model prediction
5%
25
20
4%
15
3%
10
2%
CO Cycle length
1%
5
0%
0
0
20
40
60
80
100 120
Purge gas rate , Nm /m
20%
40%
70%
100%
Vegetable oil in feed , vol%
Figure 7 Impact purge rate on CO in treat gas and cycle length (20% soybean)
Figure 6 %HDO vs vol% soybean oil in feed
78
PTQ Q3 2023
www.digitalrefining.com
Powered by FlippingBook