Appendices

79
 Distillation Procedure
A
The distillation procedure described in this appendix is based on ASTM D2892 (2005) and on experiences
gained in present study.

Prior to distillation:

• Clean and dry all glass-ware.

• Apply lubricant to all o-rings and glass-glass connections.
• Check the set-up for leakages: according to ASTM D5236 check for leaks by pumping system to 0.4
torr. After 1 min the rise in pressure should be no greater than 0.075 torr.
• Turn on cooling water: when the expected IBP is above 65◦ C cooling water at ambient temperatures
is sufficient.
• Mix dry ice and acetone in the cold trap to avoid gases to enter the vacuum pump.
• Mount the heating jacket and flask insulation.
• When running under vacuum, remember to open the reflux valve to obtain vacuum in the entire
system. If a vacuum hose is mounted to both the fractionating column and the fraction collector
this is automatically obtained.
• Turn the fraction collector tip so that the first fraction is collected in flask 1.
• Before starting distillation make sure that the system is running at total reflux.

Preparing the oil sample:

• A charge of minimum 1.2 litres is used.

• Measure the weight and volume of the oil sample.
• Add boiling stones to the distillation flask (approximately 15 g).
• The oil sample should contain less than 0.3 vol.% water. Karl-Fischer titration is used to determine
the water content of the sample. The distillation column at 100 torr is used to dehydrate the sample
if nessecary.

Initial Distillation:

• Heat up the flask at such a rate that vapours reach the top of the column between 20 and 50 min
after start-up.
• Run at total reflux until equilibrium is reached for the vapour temperature - but no longer than 15
min.
• Keep the heating jacket temperature 0-5◦ C below the vapour temperature.
• Maintain a reflux ratio of 5:1.
• A fraction size corresponding to 25◦ C in vapour temperature is used.

81
 • At the end of each fraction and at each cut point note down the time of running the distillation,
the volume of the collected fractions, the vapour and the flask temperature. Use the formulas in
Appendix B to convert ot AET.
• Do not allow the vapour temperature to exceed 210◦ C or the pot flask temperature to exceed 310◦ .
(observed temperatures)
• Continue the distillation until the desired maximum vapour temperature is reached or until the
charge shows sign on cracking. Latter being indicated by fog appearing in the flask and later at the
reflux divider. When the maximum vapour temperature is reached shut off the reflux valve, the
heating mantle and jacket. Then let the system cool down to a temperature that allows the pressure
to be lowered to the next desired value without causing the oil sample to boil.
• Note: Flush the system set-up with nitrogen, do not use air due to risk of explosion when hot.
• Weigh and collect all fractions and then reinstall all flasks.

Distillation at lower pressures:

• Turn on heating and re-establish reflux at any moderate rate for about 15 min to reheat the column
to operating temperature.
• Lower reflux ratios are allowed at lower pressures approaching 2:1 at 10 torr.
• If wax formation and crystallisation occur at lower pressures, increase the coolant temperature by
using a stream of air.
• When the final cut point is reached or when further distillation is prevented by maximum
temperature being reached follow the procedure as described in ’Initial Distillation’.

82
 Considerations Regarding
the ASTM Distillation
Guidelines
B
This appendix presents considerations related to fractional distillation and operating pressure. When
operating at different vacuum pressures, AET is an appropiate tool which is first described.

B.1. Atmospheric Equivalent Temperature

During vacuum distillation the observed temperatures must be converted to atmospheric equivalent
temperatures. Maxwell and Bonnell (1957) derived equations to convert an actual observed temperature
obtained at sub-ambient pressure to AET. For pressures above 2 torr the correlation between observed
temperature and AET is linearised. As the pressure is decreased below 2 torr, this linearised correlation
becomes inaccurate and another correlation must be applied from 2 to 0.1 torr. The equations applied in
order to convert the observed temperatures to AET are given in Equations (B.1) through (B.3). (ASTM
D5236, 1999)

748.1 · A
AET = − 273.1 [◦ C] (B.1)
[1/(T + 273.1)] + 0.3861 · A − 0.00051606

The factor A is determined through Equation (B.2) if the pressure is above 2 torr, and Equation (B.3)
applies for pressures between 0.1 and 2 torr. Note, that the equations are only valid for pressures in torr.

5.994295 − 0.972546 · log10 · P

A= for P > 2 torr [−] (B.2)
2663.129 − 95.76 · log10 · P

6.761559 − 0.987672 · log10 · P

A= for 0.1 < P < 2 torr [−] (B.3)
3000.538 − 43.00 · log10 · P

B.2. Choice of Operating Pressure

The ASTM standard D2892 used as guideline in present study is capable of producing fractions with a
final cut temperature of 400◦ C. The standard suggests operating pressures in the range from atmospheric
to 2 torr. The vacuum pump in present set-up is able to reach an operating pressure of 0.1 torr.
Lowering the pressure to 0.1 torr enables higher boiling temperatures (ASTM D5236, 1999). However,
the maximum boiling temperature reached in present study is around 400◦ C at 0.1 torr.

The distillation in present study is initiated at 100 torr instead of at atmospheric conditions. This due
to the first fraction being collected at 100◦ C. When initiating the distillation at vacuum conditions the
risk of cracking is decreased. As described in Subsection 2.2.1, some oxygenates tend to crack at lower

83
temperatures and thus a lower pressure is believed to be advantageous in the distillation of HTL bio-
crudes. The standard does not specify at which vacuum pressures the distillation should be performed.
However, it states that only one pressure level between 100 torr and 2 torr is allowed. The intermediate
pressure should be selected according to the composition of the oil sample and the desired final cut
temperature. To avoid cracking the pot flask temperature and the vapour temperature is not to exceed
actual temperatures of 310◦ C and 210◦ C respectively (ASTM D2892, 2005). In addition, the choice of
intermediate pressure affects the distillation time. Table B.1 shows how the distillation time increases
with a decreasing operating pressure. This is because the vapour expands at lower pressures and a
lower mass flow of vapours can enter the column. As a result, the distillation rate is reduced. The table
is based on a 25 mm Propak packed distillation column, corresponding to the experimental set-up in
present study. (Montemayor, 2008)

Effect of Operating Pressure on Distillation Time for D2892.

Operating Pressure Take-off Rate Distillation Time Take-off Rate Distillation Time
at 5:1 Reflux at 2:1 Reflux
[torr] [ml/h] [min/L] [ml/h] [min/L]
760 530 110 - -
100 400 150 - -
10 150 400 300 200
2 100 600 200 300
Table B.1. The distillation time as a function of the operating pressure. (Montemayor, 2008)

The increase in distillation time must be kept in mind when lowering the operating pressure. At pressures
below 10 torr ASTM D2892 (2005) suggests a reflux ratio of 2:1. However, this decreases distillation
accuracy.

Pressure Drop Related to Operating Pressure

Specifications for the distillation set-up at 75 % of maximum boil-up and atmospheric pressure are given
in Table B.2. The boil-up rate is the quantity of vapour entering the column per unit of time. The dynamic
hold-up is the quantity of liquid held up in the column under normal operating conditions. (ASTM
D2892, 2005)

15 theoretical plate column @ 75 %

Boil-up and atm. pressure (760 torr)
Column diameter, mm 25
Packing size, mm 4
Packed height, cm 57
Packed Volume, mL 280
Boilup, mL/h x cm2 650
Dynamic holdup, mL 47
Pressure drop, torr 5.1
Table B.2. Specifications of the set-up used in present study. (ASTM D2892, 2005)

The pressure drop across the packed column increases with a decreasing operating pressure (Mon-
temayor, 2008). Based on results from a leak test of the set-up there is only mass flowing through the
packed column and not through the remaining part of the set-up. Therefore, only the column pressure
drop needs to be considered. D2892 suggests to measure both the vapour and the pot flask pressure to
obtain accurate results. As mentioned the vapour pressure is measured by a pressure transducer from

84
the top of the reflux condenser, but with the given set-up it has not been possible to measure the pot flask
pressure.

Since the correlation between column pressure drop and operating pressure is unknown for the given set-
up, a constant pressure of the set-up has been used in present study. This assumption results in a higher
vacuum in the pot flask than is actually the case and thus a too high temperature in the pot flask. The
difference between the observed pot flask and vapour temperature at different operating pressures can
be used to underline this statement. Table B.3 shows the temperature difference depending on operating
pressure. The table is based on results from the distillation of CBS44.

Operating Pressure Temp. Difference

[torr] [◦ C]
100-10 60-70
0.1 110
Table B.3. Temperature difference between the observed pot flask and vapour temperature.

At 100-10 torr the temperature difference is rather constant whereas it increases significantly at 0.1 torr.
At this operating pressure the pot flask pressure is actually higher than what is modelled. Therefore, the
AET is calculated to be higher, which gives a higher temperature difference as given in the table.

Leak Testing the Set-up

In order to obtain accurate results in the distillation it is important that the set-up is leak tested. Leakages
in the distillation column will disturb the flow of vapours and condensed vapours through the column
and thus the separation sharpness of the distillation. D2892 does not mention any method to leak test
the system. D5236 states that during a leak test the pressure is not allowed to increase more than 0.075
torr in one minute when the initial vacuum pressure is 0.4 torr (ASTM D5236, 1999). The guidelines of
D5236 is used in the leak test in present study. The D5236 standard are suited for higher vacuum than
the D2892 and thus using this leak test is considered more than satisfactory.

A leakage in the fractional collector will result in a mass flow from the fractional collector into the top
of the distillation column. The pressure difference that arises will disturb the distillation accuracy by
leading some of the vapours back to the distillation head. Furthermore, a leakage between the pot flask
and the column will result in a cold air flow entering the bottom of the distillation column. Consequently,
an increased air flow in the column, will affect the equilibrium of vapours and liquids through the
column. Leakages in the hoses and hose-connections are considered less important since they are located
between the pump and the set-up. Thus they only increase the pump power and do not affect the
separation sharpness.

The original set-up presented in Figure 2.2 is not able to fulfil the leak test guidelines of D5236. Thus high
temperature vacuum grease is applied to all connections of the set-up. Furthermore, an extra vacuum
hose is mounted from the fractional collector to the vacuum pump to ensure no air flow back to the
distillation head. The vacuum grease significantly improves the leak test. A pressure increase of 0.07
torr in one minute from an initial vacuum of 0.4 torr is sufficient to fulfil the guidelines of D5236 and
thus the set-up is not improved any further.

Procedure of Lowering Pressure

Caution must be taken during the distillation when gradually lowering the pressure. Since the boiling
point decreases, the collected fractions in the product receivers must be emptied before lowering
the pressure. By emptying the receivers the collected fractions will not evaporate as the pressure is
decreased. Furthermore, the remaining oil in the pot flask must be cooled to an extend that will not
cause the liquid to boil at the new pressure. Equations (B.1) through (B.3) can be used to determine how

85
the boiling point changes with the change in pressure. When lowering the pressure do it gradually so
that any sudden unexpected boil-up can be controlled.

Before emptying the product receivers atmospheric pressure must be reached in the distillation column.
This is achieved by venting the column with nitrogen or any other inert gas. Air cannot be used to vent
the hot set-up, since this is suspected of initiating explosions in the column (ASTM D2892, 2005).

Another guideline in the ASTM standard D2892 regarding lowering the operating pressure of the set-
up is to reheat the column to operating temperature after letting the set-up cool. The column should
be reheated by reestablishing reflux for approximately 15 min. After reheating the column, nitrogen
should be flushed through it to increase the pressure and thereby drop the holdup into the pot flask.
This guideline has however been neglected in present study.

86
 Experimental Estimation of
Tcold
C
The microbatch reactors are designed with a valve and a pressure transducer that cannot resist the high
temperatures present within the sandbath. To protect these components, this top part of the reactor,
referred to as Vcold , is located above the sandbath in ambient air. As a result, the temperature Tcold inside
Vcold is lower than the operating temperature measured inside the sandbath. If not taken into account,
this induces an error in the modelling. The error is more significant the bigger the ratio between Vcold
and reactor volume. This explains why the reactors are designed to minimise Vcold to 10 % of Vtotal .

To estimate Tcold a microbatch reactor has been purged with 100 bar of nitrogen and gradually heated
up from 200 to 370◦ C. The temperature and pressure profile during this heat up is given in Figure C.1
to the left. When temperature, pressure and both Vcold and Vtotal are known, Tcold can be calculated for
the whole temperature range. This is done using the modelling software Ees, since large amount of data
is easily imported into this software. The green line in Figure C.1 represents the estimated Tcold . Despite
an extensive amount of data points and the small jumps during heat up, Tcold is surprisingly linear.

Nitrogen Experiment Tcold as Function of Operating Temperature

400 250 140

Temperature
350 130
operating range
120
300
200 110
Pressure [bar]

250
Tcold [ oC]
Temp [ oC]

100
Temperature operating
200
range 90

150 150
80

100
Assumption:
70
Tcold = 100oC
Temperature
50 Tcold 60

Pressure 100
50
0 200 220 240 260 280 300 320 340 360 380 400
0 10 20 30 40 50 60 70 80 90 100 110
Time [min] Temp [ oC]

Figure C.1. Left: Temperature and pressure profile of nitrogen experiment. Right: Tcold as function of sandbath
temperature.

The right graph in Figure C.1 is Tcold as function of operating temperature. The dashed lines in both
graphs indicate the temperature operating range used in the deoxygenation experiments. Note how
Tcold is just below 80◦ C and just above 120◦ C at the lower and upper limit of operating temperatures
respectively. A linear approximation of Tcold could be applied in the modelling, but for simplicity an
average value of 100◦ C is chosen. This is considered of sufficient accuracy when operating in a relatively
small temperature range.

87
 Deoxygenation Procedure
D
The upgrading procedure is based on experiences gained in present study.

Prior to deoxygenation:

• Clean and dry the microbatch reactor.

• Heat up the sandbath to the desired temperature.
• Control the air flow through the sandbath. Be aware that the required flow is higher in the
beginning when the sandbath is cold. As the sand is heated up the required air flow decreases.
• Prepare a text-file in LabView to store the temperature and pressure data.
• Use the VMGsim model to decide the sample size and H2 purge pressure to obtain the desired
pressure and H2 :sample molar ratio at a given sandbath temperature.
• Prepare the bottom of the microbatch reactor with sample, glass beads and catalyst (crushed
throughout conducted experiments, except CKU11). Make sure to weigh everything. (Note:
Repeatability issues reveal that the catalyst is not to be crushed.)
• Apply grease to the bottom reactor before assembling the entire reactor.
• Leak test the reactor with nitrogen. In present study the reactor is tested until 180 bar. If leaks are
discovered, disassemble, apply grease and test again.
• Purge the reactor with hydrogen. The purge pressure is determined in VMGsim.
• Mount the pressure transducer and start collecting data in the LabView program. Make sure that
the pressure is stable before lowering the reactor into the sandbath.
• Check the frequency of the frequency controller that controls the speed of the shaker.

Deoxygenation:

• When the desired temperature of the sandbath is reached the reactor can be lowered into the bath.
Make sure that LabView is logging the temperature and pressure data.
• Lower the reactor into the sandbath. The reactor should not be fastened too close to the bottom
heaters, but the volume above the bath should also be minimised.
• Turn on the shaker at the desired speed.
• Secure pressure transducer wires from the heat.
• During deoxygenation check the air flow and temperature from time to time. Furthermore, apply
oil to the connection between the engine shaft and the steel rod to ensure smooth operation.

89
After deoxygenation:

• Cool the reactor rapidly. Keep LabView logging until a steady pressure is reached.
• Clean the outside of the reactor for water and sand.
• If a mass balance is desired:

– Measure the weight of the entire reactor before and after the gases are released. The gases can
be collected for analysis. This has however not been done in present work.
– Note: Release gases in fumehood, since e.g. H2 S are likely deoxygenation products.
– When opening the reactor and separating the products make sure to measure the weight of
all components and devises used in the separation.
• In present work the products are separated in a centrifuge at 3500 rpm/2931·g for 10 minutes. The
product is collected using a syringe after centrifugation.

90
 Experimental Deoxygenation
Results
E
This appendix summarises the temperature and pressure profiles of each deoxygenation experiment.
Likewise, pictures of feed and products are given for each experiment along with corresponding FT-IR
spectra. To simplify reading, Table 5.1 is repeated in Table E.1.

Deoxygenation Experiments
CKU2 CKU5 CKU6 CKU7 CKU8 CKU10 CKU11
Unit R1 R2 R1 R2
Reactant Frac.7 CBS44 CBS44 CBS44 CBS44 CBS48 CBS48
Bio-crude g 3.90 4.90 5.02 5.04 4.82 4.93 5.04 4.94 4.98
Catalyst % 20.0 21.9 19.8 19.9 20.0 21.9 20.8 19.3 19.3
Glass beads g 0.15 0.24 0.20 0.17 0.18 0.17 0.17 0.19 0.17
Time min 193 119 123 121 189 121 120
Temp ◦
C 345 365 330 295 330 330 330
Molar ratioA
1.7 2.0 1.9 1.9 2.0 2.0 2.0 2.0 2.0
Ppurge bar 27.7 49.5 48.6 49.5 47.9 48.9 49.7 49.0 49.2
Pmax bar 63 120 99 97 109 110 103 95 96
Pmodel bar 56 101 94 90 93 95 96 94 94
Pf inal bar 8.1 30.4 19.0 33.4 36.4 25.6 33.0 18.2 19.6
HHV kJ/g 45.14B 42.50C 43.81C 42.40C 42.76C 43.94D 43.34D 44.41D 44.34D
A
VMGsim model B
Frac.7: 42.85 [kJ/g] C
CBS44: 41.19 [kJ/g] D
CBS48: 42.11 [kJ/g]
Table E.1. Table with the operating parameters used in the deoxygenation experiments and HHV of reaction
products. CKU10 and CKU11 included use of two reactors (R1 and R2) simultaneously.

Figure E.1 illustrates the temperature and pressure profile of the deoxygenation of fraction 7 from the
distillation of CBS44. The picture of the feed and the deoxygenated product shows an evident change.
The FT-IR spectra shows an upgraded product similar to diesel.

91
 CKU2 − Effect of Shaking Stability
360 70
1 Temperature
Pressure
355 60
0.8
Transmittance

350 50

Pressure [bar]
0.6
Temp [ oC]

345 40

0.4 340 30

CBS44
0.2 335 20 F7 CKU2 F5air
∆P
F5n2
330 10
0
4000 Shaking 3500No shaking 3000
Shaking 2500 2000 1500 1000
325 0 −1
0 20 40 60 80 100 120 140 Wavenumber
160180 [cm ]
200
Time [min]
Deoxygenation, Fraction 7
1
C=O
O−H
0.8
Transmittance

C−O
0.6

0.4

Fraction 7
0.2
CKU2
Diesel
0
4000 3500 3000 2500 2000 1500 1000
−1
Wavenumber [cm ]

Figure E.1. Temperature and pressure profile along with picture and FT-IR spectra of feed and product of CKU2
deoxygenation experiment. The FT-IR spectra of CKU2 is compared to that of diesel to show similarities.

Figure E.2 shows the temperature and pressure profiles along with pictures of the feed and the products
of CKU5, CKU6 and CKU7. Operating temperatures of 365, 330 and 295◦ C are tested in CKU5, CKU6
and CKU7 respectively.

92
 CKU5
375
120

110
370
100 CKU5 CBS44

90

Pressure [bar]
365
Temp [ C]

80
o

70
360
60

50
355
40
Temperature
Pressure 30
350
0 20 40 60 80 100 120
Time [min]

CKU6
360 110

355 100

350 90 CKU6 CBS44

345 80 Pressure [bar]

340 70
Temp [ oC]

335 60

330 50

325 40

320 30

315 Temperature 20
Pressure
310 10
0 20 40 60 80 100 120 140
Time [min]

CKU7
310 100

305 90 CKU7
CBS44

300 80
Pressure [bar]
Temp [ oC]

295 70

290 60

285 50

280 40
Temperature
Pressure
275 30
0 20 40 60 80 100 120
Time [min]

Figure E.2. Temperature and pressure profile and picture of feed and products of CKU5, CKU6 and CKU7
deoxygenation experiments.

93
Figure E.3 illustrates the results of CKU8 where the residence time is prolonged from 2 to 3 hours. The
pressure and temperature profile shows repeatability issues. Based on this experiment the capacity of
the deoxygenation set-up is extended to handle two microbatch reactors which is used in CKU10 and
CKU11.

CKU6 and CKU8 − Repeatability

345 120

CBS44 CKU8
340
100
335 Deoxygenation, CBS44
1
330 80

Pressure [bar]
Temp [ oC]

0.8
325
Transmittance

60
0.6
320

315
∆P 40
0.4 ∆P
CBS44
310 CKU6, temperature
20 CKU5
0.2 CKU6, pressure
305 CKU6
CKU8, temperature
CKU8, pressure CKU7
0
300 0
40000 20 40 3500
60 80 100 3000
120 140 160 2500 200
180 2000 1500 1000
Time [min] Wavenumber [cm−1]

Deoxygenation, CBS44
1

0.8
Transmittance

0.6

0.4

CBS44
0.2
CKU6
CKU8
0
4000 3500 3000 2500 2000 1500 1000
Wavenumber [cm−1]

Figure E.3. Temperature and pressure profile, picture and FT-IR spectra of feed and product of CKU8 deoxygenation
experiment. The results are compared to CKU6 to visualise repeatability issues.

94
 CKU10 − Repeatability
340 120

338 110

336 100
R1 R2
CBS48
334 90

Pressure [bar]
332 80
Temp [ C]
o

330 70

328 60

326 50
∆P
324 40
Temperature ∆P
322 Pressure R1 30
Pressure R2
320 20
0 20 40 60 80 100 120
Time [min]
Repeatability, CBS48
1

0.8
Transmittance

0.6

0.4

Feed
0.2
CKU10, R1
CKU10, R2
0
4000 3500 3000 2500 2000 1500 1000
Wavenumber [cm−1]

Figure E.4. Temperature and pressure Repeatability,

profile, pictureno
and FT-IR spectra
crushing, CBS48of feed and products of CKU10
1 deoxygenation experiment.

0.8
Transmittance

CKU10
0.6 and CKU11 are experiments investigating the repeatability issue discovered in CKU8. CKU10
shows that repeatability is not obtained when crushing the catalyst, whereas CKU11, where the catalyst
is not
0.4 crushed, shows similar pressure profiles in the two reactors.

Feed
0.2
CKU11, R1
CKU11, R2
0
4000 3500 3000 2500 2000 1500 1000
Wavenumber [cm−1]

95
 CKU11 − Repeatability
340 100

335 Repeatability,
90 CBS48
1
330 80
R1 R2
0.8 CBS48
325 70
Transmittance

Pressure [bar]
Temp [ oC]

320
0.6 60

315 50
0.4
310 40

0.2 ∆P Feed
305 30 CKU10, R1
∆P
Temperature CKU10, R2
300
0 Pressure R1 20
4000 Pressure3500
R2 3000 2500 2000 1500 1000
295
0 20 40 60 80 100 120
10
Wavenumber [cm−1]
Time [min]
Repeatability, no crushing, CBS48
1

O−H
0.8
Transmittance

C=O C−O
0.6

0.4
Feed
CKU11, R1
0.2
CKU11, R2
Shell Crude−Oil
0
4000 3500 3000 2500 2000 1500 1000
−1
Wavenumber [cm ]

Figure E.5. Temperature and pressure profile, picture and FT-IR spectra of feed and products of CKU11
deoxygenation experiment.

96