ENG60 LNG II ENV60 – LNG II Metrological support for LNG custody transfer and transport fuel applications Report for the validation of the calculation of the Methane Number and correlation to the LNG composition Task 3.4, deliverable 3.4.3: Report on the validation of a calculation for the MN from the LNG composition and the correlation of the MN to the LNG composition (VSL, NPL, PTB, TUBS (REG))
26
Embed
ENV60 LNG II Metrological support for LNG custody transfer ...€¦ · ENG60 LNG II ENV60 – LNG II Metrological support for LNG custody transfer and transport fuel applications
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
ENG60 LNG II
ENV60 – LNG II
Metrological support for LNG custody
transfer and transport fuel applications
Report for the validation of the calculation of the
Methane Number and correlation to the LNG
composition
Task 3.4, deliverable 3.4.3: Report on the validation of a calculation for the MN from
the LNG composition and the correlation of the MN to the LNG composition (VSL, NPL,
PTB, TUBS (REG))
ENG60 LNG II
TABLE OF CONTENTS 1. Introduction ......................................................................................... 3
2. Literature ............................................................................................ 4
2.1 Definition Methane Number ................................................................. 4
Ignition delay times of different reference mixtures in the range of MN60 to
MN100 were measured at various temperatures and pressures (20 and 40 bar) at
a stoichiometry of 𝜙=0.4. Therefore, the reference mixtures were premixed with
oxygen and diluent gases (nitrogen/argon). An overview about the obtained
results is given in Figure 2. In the latter, the data points are fitted linear to
highlight the given trend in the ignition delay times. Please note, that the linear
fit doesn’t reflect the real situation in the RCM because it doesn’t consider the
heat loss over the walls. It is obvious that with increasing temperature the
ignition delay time becomes shorter for each reference mixture. However, a clear
trend in the behavior of the ignition delay times between the different MN
mixtures is not observable.
ENG60 LNG II
Figure 2 : Measured ignition delay times (IDT) for the reference mixtures at 20 bar (left) and 40 bar (right). The data points are fitted with a linear fit routine not considering the experimental heat loss.
Since a clear scientific picture could not be drawn from the obtained data, exemplarily kinetic simulations for the 40 bar case were performed. Modeling was carried out using the Flame Master software15 with a reduced version of the
n-heptane kinetic mechanism from Lawrence Livermore National Laboratory (LLNL)16, including 1083 reactions and 207 chemical species. The results are
shown in Fig. 4 for the fuel-lean case with an equivalence ratio of 𝜙=0.4 (left)
and for the fuel-rich case of 𝜙=1.2 (right). In the simulations heat loss, like it
occurs in the Rapid Compression Machine (RCM) experiments, was not considered, but this fact should not change the general features that can be seen
from the simulated results and that can partially explain the observed
experimental behaviour.
Figure 3 Chemical kinetic simulations of the reference mixtures’ ignition delay times at
40 bar with fuel-lean (𝜙=0.4) [left] and fuel-rich (𝜙=1.2) [right] conditions. The blue
rectangle indicates the optimal working range of the Rapid Compression Machine (RCM).The light patterned region shows the range of the experimental obtained values.
ENG60 LNG II
Figure 4 shows the simulated IDTs for the different reference mixtures in a wide temperature range for the fuel-lean case (left panel of Fig. 4)) and the fuel-rich case (right panel). The blue rectangles indicate the optimal working range of the
Rapid Compression Machine whereas the light patterned region describes the range of the results obtained in the experiments. Outside of the blue box, i.e.
outside of the optimal working range of the RCM, uncertainties of the experimental values become larger than ±15%. It should be noted that only the temperature is the controllable experimental variable (x-axis) and that the
Ignition delay time is the obtained result (y-axis). From Fig. 4 it becomes clear that in the optimal working range of the RCM the reference mixtures show similar
ignition delay times, which is consistent with the experimental observations shown in Fig. 3. As already mentioned above, heat loss was not considered in the
simulations, shifting the modelled results to lower temperatures in comparison to the experimental range (which strengthen our argumentation that IDTs are very similar for the different reference mixtures). It should be pointed out, that
ignition of MN0 and MN100 (and within the other reference mixtures) belongs to aninterplay of different chemical reactions depending on the amount of hydrogen
and methane in the mixture. However, a clear statement would require a detailed investigation by kinetic modelling including sensitivity and rate of production analyses.
INVESTIGATION OF LNG-MIXTURES
To study the knocking propensity of different LNG mixtures, a set of different mixtures, defined in Table 1 was investigated. The LNG mixtures were premixed
with oxygen and constant amount of diluent gases (nitrogen/argon) allowing
measurements in a similar temperature range at an equivalence ratio of
𝜙=0.4/𝝀=2.5. The measurements were extended to an equivalence ratio
𝜙=1.2/𝝀=0.83 covering also a fuel-rich situation and thus a deeper insight in the
ignition behaviour of LNG mixtures. Measurements were performed each at a minimum of 3 different temperatures within a temperature range from ~870-
1100 K. Ignition delay times were determined for two different pressures (20 bar and 40 bar). Each test series (LNG-Mix) was repeated three times to minimize
the uncertainty of the experimental data and averaged to extract the value for the ignition delay time. Figure 4 gives an overview of the measured ignition delay times in comparison with the reference mixtures. Due to the fact, that the
diluents contain always the same ratio of nitrogen and argon (and the mixture’s heat capacity belongs then only on its composition), some mixtures showed
ignition delay also at temperatures outside of the temperature ranges shown in Fig. 5. These measurements were not included in the figure to allow for a proper
comparison between the different mixtures. A full dataset including all averaged measured IDTs can be found in Table 2.
ENG60 LNG II
Figure 4 Determined ignition delay times for the LNG-mixtures at two different pressures and with two different stoichiometries. The bars’ colour code indicates similar temperatures
As already shown for the reference mixtures, the expected trend that with lower
temperature the ignition delay times of each mixture increases is observable.
However, a clear correlation between the mixtures’ determined methane number
(see Table 1) and the measured ignition delay times cannot be found. This is
consistent with the findings of the ignition behaviour of the reference mixtures as
discussed above. Nevertheless, it can be observed that (in the fuel-lean case)
the ignition delay times of the LNG-mixtures are shorter than for the reference
mixtures for temperatures between 964K and 1009 K (20 bar case) and 921 K
and 979 K (40 bar case), respectively. This behaviour, which is even more
pronounced at higher pressures, indicates strong influence of the other chemical
components, i.e., the higher hydrocarbons, on the ignition and thus on the
knocking resistance of the different LNG-Mixtures. However, from a chemical
point of view, ignition delay times belong strongly to the present radical pool and
the importance of radical forming and consuming reactions. A more detailed
ENG60 LNG II
analysis of a possible correlation between the MN and the ignition delay time
thus require a deeper insight into the combustion chemistry of LNG and
reference mixtures. It is proposed to develop a chemical kinetic model that can
accurately predict the combustion and ignition behaviour of LNG with different
MNs and that can help to understand the observations made during this project.
ENG60 LNG II
Table 2: Overview of all measured averaged ignition delay times for the different LNG-mixtures.
P = 40 bar, phi = 0.4
1/T (1/K) Temperature (K) IDT (ms) 1/T (1/K) Temperature (K) IDT (ms) 1/T (1/K) Temperature (K) IDT (ms) 1/T (1/K) Temperature (K) IDT (ms)
ALGORITHM Since the newly developed algorithm is based on the experimental data gathered
by AVL,1 it has been initially validated by comparing it with other popular
methods based on the AVL data namely the MWM and the Danish Gas Company
(DGC) methods. A detailed version of this comparison together with descriptions
of the different methods can be found in the document describing the developed
algorithm that has been submitted together with the algorithm.14
In summary, it was found that the MN values calculated with the new algorithm
agree well with the values derived by the MWM and the DGC methods. Figure 9a
shows calculated MN using the different methods for of a set of exemplary LNG
mixtures. Further analysis indicates that the results of the new algorithm
correlate approximately linearly with the other two methods (Figure 9 b)
Figure 9. Calculated methane numbers for a set of LNG mixtures with the DGC method (center) and the MWM method (right) plotted against the corresponding values
calculated with the new algorithm fit.
RESULTS The final uncertainty budget has validated with the input from D3.3.1, D3.3.2
and D3.4.1. Resulting values for the expanded uncertainty of the methane
number strongly depend on the composition of the LNG mixture, and are
typically between 0.3% and 0.8% relatively .
ENG60 LNG II
6. CONCLUSION A new algorithm, based on existing algorithms such as the AVL and MWM, has
been developed for LNG type of chemical composition. The new algorithm pays
special attention to the influence of higher hydrocarbons such as butanes and
pentanes as these components have a high impact on the methane number while
the concentration of these components in an amount of stored LNG can increase
over time (ageing effect). The algorithm is programmed in a user-friendly MS
Excel spreadsheet and an expanded uncertainty on the result is calculated to be
typically between 0.3% and 0.8% relative on the MN. This is the first algorithm
to provide a MN accompanied by an uncertainty.
Experimental results on two types of four-stroke single cylinder engines show a
good correlation between the calculated MN and the measured service MN.
However, the correlated line is slightly “tilted”, showing a lower SMN at the low
MN range and a higher SMN in the high range. More experiments are needed to
understand this behaviour and/or to modify the algorithm(s).
Ignition delay time measurements using a Rapid Compression Machine show
good correlation of the results with the hydrogen in methane mixtures. However,
the results with the more complex natural gas mixtures do not show the
expected correlation. The complex chemistry of the LNG related combustion
process has to be better understood. New experiments, with less complex
mixtures, such as e.g. propane in methane and butane in methane, have to be
performed to describe an improved kinetic model for LNG combustion.
The algorithm, the experimental results and the content of the report will be
communicated to relevant standardization committees to be used in the creation
of a harmonized method for the calculation of the MN.
ENG60 LNG II
LITERATURE 1. Leiker, M., Christoph, K. & Al., E. Evaluation of Anti Knocking property of
Gaseous Fuels by Means of Methane Number and its Practical Application to
Gas Engines. ASME Publ. (1971).
2. prEN 16726:2014 Gas Infrastructure- Quality of gas- Groep H.
3. POSITION, E. Methane number as Paramete for Gas Quality Specifications.
(2012).
4. Bottino, G. et al. EMRP METROLOGY FOR LNG FEEDBACK.ppt. (2014).
5. Gas Infrastructure Europe, G. Position paper on Impact of including
Methane Number in the European Standard for Natural Gas. (2012).
6. Standard Test Method for Motor Octane Number of Spark-Ignition Engine
Fuel- ASTM D2700.
7. ISO\TR 22302 Natural gas- calculation of Methane Number. (2014).
8. ISO 15403-1:2006: Natural Gas- Natural Gas for use as a compressed fuel
for vehicles part 1: Designation of the quality. (2006).
9. MN calculator. Available at: http://www.cumminswestport.com/fuel-
quality-calculator.
10. DNV GL methane number calculator. Available at: