LGI injection rate
Auteurs
S. Winkelman
Address
a DaVinci Laboratory Solutions
Sydneystraat 5, 3047 BP Rotterdam
ARTIKEL INFO
ABSTRACT
Keywords:
Internal document for the LGIvolumetric injection rate
Davinci Laboratory Solutions uses an automotive injection valve integrated with a
needle mounted on a chromatograph (GC) to inject high pressure liquefied gas named:
LGI (Liquefied gas injection). In this paper two experiments are described. One
focusses on the effects of volume injection with GC system back pressure. By injecting
into a closed system pressurized by nitrogen a backpressure effect is simulated. The
results are presented in Table 1: GC backpressure effect and a conclusion can be made
that backpressure does not affect the volume injected into the system. The second
experiment focusses on the repeatability and reproducibility of the injector/GC needle
valve combination. This done by injecting into a vial and weighing the amount
injected. The results are presented in Table 2: Repeatability: Volumetric injection LGI
& Table 3: Reproducibility: experiments. After applying statistics described in the
ASTM D7756 norm it is concluded that: That the injector-needle combination with a
99% certainty a volumetric injection presented in Appendix 4 with a 95% probability
that these values interchangeable between injectors. Hereby effectively reporting the
repeatability and reproducibility of the product.
1 Introduction
The company Davinci Laboratory Solutions (Davinci-LS) has used
an automotive injection valve to automatically inject a high
pressure sample directly into a gas chromatograph (GC) called LGI
(Liquefied gas injection). The system uses a Hitachi FSI injector
rated up to a pressure of 80 bar common rail pressure. The
injector has been modified to fit a custom GC needle effectively
creating a high-pressure injection sample system to inject liquid
high pressure samples. This system can used to inject samples
form a distillation tower or oil well to measure a representative
pressurized process sample in a GC.
To laboratory technicians its valuable to know how much
and how repeatable the amount injected is. The rate of injection
of a FSI injector has been widely modelled in literature [1] [2] [3].
The rate of injection is dependent on the speed of sound in the
fluid and the injection orifice explained by J. T. Menucci. The
mathematical model is described by the following variables:
Common rail pressure, cylinder head pressure, Injector orifice
resistance and time [2]:
𝑑(𝑚) 𝐴
= ∗ 𝑃 [2]
𝑑𝑡
𝑎
Where m is in kilogram injected mass, t in milliseconds open/close
time, A in m2 as injection orifice, a in m/s as the speed of sound
trough the liquefied gas and P in bar as common rail pressure (gas
sample bomb pressure). The speed of sound in the formula is used
because a sudden mechanical movement of fluid under pressure is
governed by the sound of sound through the fluid [1].
This short paper focuses on modelling the injection rate of the FSIcustom needle combination. This will be done by looking at the
effects of GC backpressure (cylinder head pressure) and the
overall injection rate of the system. The results are statistically
analysed for the repeatability and reproducibility plus the
mathematical model is validated.
2 Experiments
Both experiments are based on the work done by J. T. Menucci and
are adapted to fit the application [2].
2.1 GC backpressure effects
To simulate the effects of backpressure on the injection needle the
needle must inject into a closed system under a specific GC inlet
pressure ranging from 2 – 6 bar absolute. This is done by welding
a GC inlet onto a 1/16-inch tubing with a fixed geometry and
bending the tube 90 degrees and then filling the fixed geometry
with octane so that the injection is done in a fluid which equally
divides the pressure in the system. By using a 6-way valve with a
small internal volume the system can be pressurized with N2 and
closed when the desired pressure is reached. See Appendix 1 for a
piping & instrumentation diagram (PID), Custom GC inlet and
automotive adapter. When injecting a fluid in a closed system
results in a pressure increase and then the injected volume can be
calculated by using the ideal gas law:
𝑃1 ∗ 𝑉1 = 𝑃2 ∗ 𝑉2
𝑉𝑔𝑠 ∗ 𝑃0
𝑉𝑔𝑠−𝑛𝑑𝑙𝑒 =
𝑃𝑛𝑙𝑑𝑒
𝑉𝑔𝑠−𝑛𝑑𝑙𝑒 ∗ 𝑃𝑛𝑑𝑙𝑒
𝑉𝑔𝑠−𝑛𝑑𝑙𝑒−𝑖𝑛𝑗 =
𝑃1
𝑉𝑖𝑛𝑗 = 𝑉𝑔𝑠 − 𝑉𝑔𝑠−𝑛𝑑𝑙𝑒 − 𝑉𝑔𝑠−𝑛𝑑𝑙𝑒−𝑖𝑛𝑗
Subsidize and eliminate.
𝑃0
𝑃0
𝑉𝑖𝑛𝑗 = 𝑉𝑔𝑠 ∗ (
− )
𝑃𝑛𝑑𝑙𝑒 𝑃1
Figure 1: schematic of setup
injecting multiple times, the mass will stepwise increase, then by
subtracting each value the mass the single injection mass will be
found. See Appendix 2 for the PID and code.
Materials used:
Swagelok Port connectors.
Swagelok male female connectors.
Swagelok bolts.
Swagelok 1/16 inch tubing.
Swagelok butterfly valve.
Sample Bomb (100 bar).
FSI (Hitachi): 06E 906 036AE with GC needle.
100 Bar N2.
Octane stock.
Injector driver module.
Analytical scale. KERN ARS 120-4.
Swagelok EN 837-1
PC with C+ code and serial connection to injection
module and scale.
2.3 Method used
Described in paragraph 2.3.1 and 2.3.1 are the procedure for the
experiments.
2.3.1
Materials used:
Swagelok Port connectors.
Swagelok male female connectors.
Swagelok bolts.
Swagelok 1/16-inch tubing.
Swagelok butterfly valve.
Swagelok needle valve.
Sample Bomb (100 bar).
Omega Pressure sensor Model No: PXM319-200G10V
with custom digital readout using a We!ntek MT8050iE
Human Machine Interface.
FSI (Hitachi): 06E 906 036AE with GC needle.
Injector driver module.
100 Bar N2.
Octane stock.
Septum + custom made GC septum inlet.
6 way valve.
2.2 Injection amount
The experiment is done by injecting into an empty vial that is
placed on an analytical scale. Code is written in C+ to automatically
inject a sample and log the analytical scale value into a .CSV file. By
GC backpressure
Variable: Pressure 1,2,3,4,6 bar absolute measurement tube & 10
milliseconds injection time & 20, 40, 60, 80 bar Sample Bomb
pressure.
1. Fill the Sample Bomb with 40 ml octane.
2. Pressurize the Sample bomb with 100 bar N2 and
connect to the system.
3. Open sample bomb valves and spool the system.
4. Open the GC inlet.
5. Fill GC inlet with octane up to the brim and shake air
bubbles out of the system.
6. Close GC inlet with septum.
7. Turn 6-way valve into fill configuration.
8. Pressurize the system.
9. Turn the 6-way valve into measurement configuration.
10. Measure pressure.
11. Inject GC needle into GC inlet.
12. Measure pressure
13. Injection.
14. Measure pressure
2.3.2
Overall rate of injection
Variable: 25,50 milliseconds injection time & 20, 40, 60, 80 bar
Sample Bomb pressure, FSI (Hitachi).
1. Fill the Sample Bomb with 40 ml octane.
2. Pressurize the Sample bomb with 100 bar N2 and
connect to the system.
3. Open sample bomb valves and flush the system.
4. Place setup above vial on scale.
5. Connect serial connection to PC.
6. Run code.
7. Depressurize the sample bomb pressure to next value.
8. Repeat till 20 bar.
9. Swap FSI for another one.
3 Result
ANOVA statistics
3.1 GC backpressure effect
H0 = P(0,05)
F
F crit
P-value
dF
20 bar, 50ms
190,7
3,1
0,000
81
40 bar, 50ms
19,2
3,1
0,000
77
60 bar, 50ms
0,0
3,1
0,996
80
80 bar, 50ms
124,4
3,1
0,000
75
20 bar, 25ms
3,1
3,1
0,051
83
40 bar, 25ms
66,9
3,1
0,000
86
60 bar, 25ms
88,2
3,1
0,000
86
80 bar, 25ms
89,6
3,1
0,000
84
After completing the experiment, the data is calculated and plotted
in the graph below. The backpressure influences the injection
volume expected was that the injection volume decreased if
pressure increased but as shown below this is not the case.
DATA, Back pressure effect
V(uL)
5,0
4,0
20 bar
3,0
40 bar
2,0
60 bar
1,0
80 bar
Table 3: Reproducibility: experiments
3.3 Model validation
0,0
0
2
4
6
8
P(bar) relative back pressure
Table 1: GC backpressure effect
Using the physical values below and experimental values
displayed in Appendix 4 a correlation graph is made with an
ANOVA analysis with a H0 = P(0.05):
Table 4: Physical value's
Physical value's
After completing the experiments, the data is processed in excel
with suggested statistics from the product current ASTM D7756
norm form the LGI-system, see Appendix 3 for the raw data output.
After calculating the mass measured per single injection statistics
is applied using t-student statistics to calculate the upper and
lower limit of injection with a 99% certainty effectively displaying
the repeatability of the injector with a GC needle. See Appendix 4
for the actual values.
Spread with 99% confidence interval
100
a.octane(m/s)
1171
A.needle(m2)
4,9E-08
model vs experiment
100
Lineair
(m.model,
50ms)
Lineair
(m.experime
ntal, 50ms)
Lineair
(m.model,
25ms)
Lineair
(m.experime
ntal, 25ms)
80
P (bar)
3.2 Overall rate of injection
60
40
p(bar)
20
0
0,0E+00
50
1,0E-05
2,0E-05
mass (kg)
Table 5: model vs experiment
0
0,0
5,0
10,0
15,0
50ms, mu1
V(uL)
25ms, mu1
20,0
25,0
50ms, mu2
25ms, mu2
Table 2: Repeatability: Volumetric injection LGI
H0 = 0,05
F
F crit
P-value
dF
25 ms
0,29
5,98
0,60
7
50 ms
0,86
5,98
0,38
7
Table 6: model vs experiment
The FSI is swapped out 3 times to perform the same experiment
where after that the limit values are compared using a single side
analysis of variance (ANOVA) statistics effectively displaying the
reproducibility of the injector with a GC needle with a H0 =P(0.05)
accepted variation. As seen below the hypnotises can be accepted
in multiple cases, so the injector with GC needle produces
statistically reproducible volume injections. Why the F value is in
some cases higher than the F crit value will be discussed in section
4.
Looking at the F values there can be concluded that the model has
validation and the 50 ms hypothesis is accepted while 25 ms is
rejected.
4 Discussion
The experiments described in section 3.2 are done by manually
venting N2 from 80 bar to 20 bar with a manometer. A discussion
can be made about the fact that the pressure was measured
analogously and was therefore not exactly at the setpoint ranging
from 80 till 20 bar. Therefore, the F values presented in Table 3
can be explained, the injection volume is strongly dependent on
the gas sample bomb pressure and the volumes will thus not me
the same if the pressure is slightly different. The same explanation
can be made for the P values of table 6.
6 Appendix
6.1 Appendix 1
5 Conclusion
A system designed by Davinci-LS uses an automotive injector
valve. This valve is modified with a gas chromatograph needle to
be used as high-pressure liquefied gas injection into a gas
chromatograph. An experiment has been setup to measure the
effects of system back pressure and injection rate of the system.
After reviewing the back pressure experiment there can be
concluded that backpressure does not affect the injection volume.
After preforming the rate of injection experiment 3 times that by
injecting though a custom needle, swapping the injector and
performing statistics on the result a conclusion can be made that:
with 99% certainty the injected volume octane is as is presented
in table 2 and Appendix 4 effectively displaying the repeatability
of the product. Then the reproducibility is calculated, aside from
value’s diverting due to a gas bomb pressure setpoint deviation
(as described in the discussion) the hypothesis is accepted that
95% of an x number of injections lay between the values shown in
table 2 and Appendix 4.
After comparing the model values with the experimental values a
deviation ranging from 50% to 5% is found between the two. The
cause of this variation is due to the seen leakage of the injector
when the custom needle is attached. Further research is needed
why the injector leaks when the needle is attached.
6. Bibliography
Figure 2: PID setup 1
[1] W. Bosch, “The Fuel rate Indicator: A new measuring Instrument For Display of
the Characteristics of Individual Injection,” SAE International, USA, 1967.
[2] T. J. Menucci, “Development of Bosch Rate of Injection measurement Procedure
and Results,” Michigan Technological university, Michigan, 2018.
[3] M. Shashank, “Piezoelectric Diesel Injectors & Emission Control,” International
Journal of Science and Research, India, 2015.
[4] Bosch, “HDEV6: Injection pressures up to 350 bar,” 2018. [Online]. Avialable:
https://www.bosch-mobility-solutions.com/en/solutions/valves/highpressure-injector/.
Figure 3: Custom GC inlet with septum and pressure sensor
Figure 4: Injector plus custom needle adapter
6.2 Appendix 2
Figure 6: injection Module, scale serial connection
Figure 5: PID setup 2
Figure 7: Experimental setup
6.3 Appendix 3
Table 7: Raw data
50ms injectie
0,5
0,45
0,4
mass(g)
0,35
0,3
0,25
0,2
0,15
0,1
0,05
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29
injection counter
20bar, 25ms gewicht
40bar, 25ms mass
60bar, 25ms mass
80bar, 25ms mass
Table 8: Raw data
25ms injectie
0,25
mass (g)
0,2
0,15
0,1
0,05
0
1 3 5 7 9 11 13 15 17 19 21 23 25 27 29
injection counter
20bar, 25ms gewicht
40bar, 25ms mass
60bar, 25ms mass
80bar, 25ms mass
6.4 Appendix 4
t-student
V.limits (uL)
20 bar, 50ms
11,0
10,4
40 bar, 50ms
16,8
16,4
60 bar, 50ms
20,5
20,2
80 bar, 50ms
23,4
23,1
20 bar, 25ms
6,4
5,9
40 bar, 25ms
9,6
9,1
60 bar, 25ms
11,3
11,0
80 bar, 25ms
12,8
12,4