The Master Chemical Mechanism
The following exercises look at the modelling of smog chamber measurements of ethene oxidation carried out at EUPHORE as part of the EXACT campaign in 2001.
A controversial issue regarding the evaluation of atmospheric mechanisms through the use of photo-smog chamber data is the influence of chamber dependent reactions on the system being studied. Such chamber dependent processes include: the introduction of free radicals from heterogeneous wall reactions, adsorption/desorption of NOy species to/from the walls and off-gassing of VOC species from the walls which contribute to ozone formation. Obviously, such "background" chamber dependent mechanisms are chamber specific.
Other important chamber processes that need to be taken into account include dilution of reactants and products (via replenishment of gases sampled or lost from the chamber) and, for outdoor environmental chamber, the manner in which solar actinic flux is modified via transmission through the transparent chamber walls and subsequent albedo effects of the chamber floor.
In order to gain insight into the chamber effects at EUPHORE, two ethene photo smog experiments were carried out as part of the EXACT campaign in 2001. The experiments carried out on the 11/09/2001 and 01/10/2001 were performed under initial "low NOx" (1:25) and "High NOx" (1:3) conditions respectively. Table 1 shows the initial conditions and other important parameters for the two experiments. Ethene was chosen because it is a simple VOC whose photo-oxidation mechanism is well known.
For more information see Bloss et al. (2005a and b) and Zador et al. (2005)
Table 1. Initial concentrations and other parameters for the two EXACT ethene experiments
|11/09/2001 (low NOx)||01/10/2001 (high NOx)|
|Start time (hh:mm)||10:00||10:05|
|End time (hh:mm)||15:00||16:00|
|H2O (ppbv)||6.4 ×10-5||3.8 × 10-5|
|Dilution Rate (s-1)||1.34×10-5||1.64 ×10-5|
Locate the ethene mechanism on the MCM website. Extract the ethene mechanism using the "subset mechanism extractor".
Paste the mechanism into the appropriate place in the model eth_011001.fac .
Paste the species VARIABLES and RO2 summation list into the appropriate place in the model. (NOTE: As already mentioned, take care that the extracted summation list does not contain any Criegee Intermediates (radicals whose name will end in “OO”, “OOA” or “OOB” etc…)). Also, some of the inorganic species are repeated in the extracted species VARIABLES list. Remove these.
Initialise the model to the initial values from the high NOx experiment carried out on the 01/10/2001 (Table 1), except for the dilution rate. For the photolysis rate calculations make sure you define the date of the experiments. Start the model at the appropriate time and output every 5 minutes until the end of the experiment (72 times).
Run the model and compare the results with the experimental data contained in eth011001.xls by pasting your results into the “model1” worksheet.
Q1. How well are the concentration profiles simulated?
As already stated, these experiments feature a slow dilution of the reactants and products in the chamber through small leaks due to the inner pressure of the chamber being marginally higher than ambient, which prevents contamination by inflow of ambient air. This loss is made good by adding clean air to the chamber during the experiment.
In order to measure the dilution rate an inert tracer, in this case SF6, is added at the beginning of the experiment and its concentration is monitored over time by FTIR. The calculated average first order loss rate of SF6 is used as the dilution rate and is applied to all stable species in the model (these loss reactions are listed after the VOC mechanism in the chamber models).
Add the measured SF6 dilution rate as listed in Table 1 into the INITIAL routine of the model.
Re-run the model and compare the results to the experiments and the first run by pasting your results into the "model2" worksheet in eth011001.xls.
Q2. How do the new concentration profiles compare to the measurements and the model without a dilution rate?
The effects of the chamber dependent reactions at EUPHORE were investigated using the results of the two ethene experiments. A base case auxiliary mechanism was constructed from EUPHORE characterisation experiments and literature data adapted to EUPHORE conditions (Bloss et al. 2005a). Discrepancies between the modelled and measured data and a detailed sensitivity analysis were used to derive a tuned auxiliary mechanism which is listed in Table 2.
Table 2. Parameters from the tuned auxiliary mechanism used to assess the impact of chamber related processes on the ethene experiments
|Process||Tuned (using both experiments)|
|NO2 = HONO||0.7 ×10-5s-1|
|NO2= wHNO3||1.6 × 0-5s -1|
|O3= wO3||3.0 × 10-6s-1|
|Initial HONO||NOx dependent|
Add the EUPHORE tuned auxiliary mechanism to the end of the reaction listings in EQUATIONS. Remember to declare any new VARIABLES.
Re-run the model and compare the results to the experiments and the other model runs by pasting your results into the "model3" worksheet in eth011001.xls.
Q3. How do the new concentration profiles compare to the measurements and the other model runs?
All the calculated photolysis processes apply scaling factors in order to take into account the transmission through the walls, backscatter from the aluminium chamber floor and cloud cover.
The variable scaling factor F1 is applied to all photolysis rates in order to take into account wall and cloud transmission effects and will be dependent on the species absorption cross-section. In a previous EUPHORE experiment the photolysis rates of NO2 (j(NO2)), O3 (j(O1D)), HCHO (j(HCHO))and HONO (j(HONO)) were measured. For these species the F1 scaling factors are based on the deviation between their measured and calculated photolysis rates, normalised to the deviation seen for j(NO2). For all other photolysis rates the average value of these factors is used. The F1 scaling factors are applied to the photolysis rate parameterisations in the model.
j(NO2) is routinely measured in chamber A at EUPHORE and these data are available for both experiments. Variations in actinic flux from day to day and during the experiment resulting to short temporal scale variations in cloud cover are account for by considering the difference between the measured and clear sky calculated j(NO2) at any given time during the experiment. This variable scaling factor JFAC is applied to all calculated photolysis rates along with F1.
In order to constrain the model with measured values of j(NO2) save the file jno2.in to the directory where you are running your model from. "jno2.in" contains measured j(NO2) values averaged to the appropriate time steps of the model.
NOTE: When you constrain a VARIABLE species you will have to remove it from the VARIABLES declarations list and declare it as a PARAMETER.
Enable the model to read in these measured values of j(NO2). At the beginning of the model open "jno2.in" so the model can read data from it.
Next declare the array "JNO2IN" to read the data into as a PARAMETER and make sure the array size is correctly defined.
Read values from "jno2.in" into "JNO2IN" in the INITIAL routine.
In the DATAIN routine make the data in array "JNO2IN" equal to "J<4>". "J<4>" is defined as j(NO2) in the MCM.
In the photolysis rate calculation section enable the "JFAC" equation. This equation calculates the scaling factor due to changing cloud cover.
Finally, comment out the photolysis rate parameterisation for j(NO2).
IMPORTANT!: ALL input data files need to be in the same file as the model is run from.
Save and re-run the model and compare the results to the experiments and the other model runs by pasting your results into the "model4" worksheet in eth011001.xls.
Q4. How do the new j(NO2) constrained model concentration profiles compare to the measurements and the other model runs?
Initialise your new j(NO2) constrained model containing the EUPHORE tuned auxiliary mechanism to the initial values from the low NOx experiment carried out on the 11/09/2001 (Table 1). Remember to start the model at the appropriate time and output every 5 minutes until the end of the experiment. Also constrain with the appropriate experimental values of j(NO2) click here for jno2.in for 11\09\01
Save the new model as eth_110901.fac.
Run the model and compare the results with the experimental data contained in eth110901.xls by pasting your results into the "model1" worksheet.
The model can be used to analyse the chemical system. You can perform rate of production (and destruction) analyses on any species created and destroyed in the system.
The following exercise looks at a simplified ROPA for HO2 in the 01/10/2001 ethene experiment.
Open the model eth_011001_ropa.fac
We could calculate the rates of production and destruction by defining a series of equations, for example:
ROP1 = 1.90D-12*EXP(-1000/TEMP)*OH*O3
ROP2 = …
ROD1 = 1.40D-14*EXP(-600/TEMP)*HO2*O3
ROD2 = …
However, if you look through the eth011001_ropa.fac you will see that each reaction has a G and S value associated with it. The G values give the instantaneous rate, which is what we have defined in the above equations, and the S values the cumulative rate for each reaction.
• Identify the G numbers for the reactions that produce and consume HO2 in the ethene model.
As you have noted there are many reactions that can create and destroy HO2 in the ethene system. However, only a relatively small number will be important under the conditions of the experiment. To save time, the five most important reactions with respect to HO2 production and destruction are listed in Table 3.
Table 3. Model G values of the five most important HO2 formation and destruction reactions for ethene under the conditions of the experiment carried out on the 01/10/2001.
|HO2 Production||HO2 Destruction|
• Modify the output instructions in eth011001_ropa.fac in order to calculate the G values for HO2 production and destruction listed in Table 3.
Q5. Run the model and plot the results in a spreadsheet: You can see the reaction rates for each reaction in Table 3 producing and consuming HO2 and how they change throughout the course of the experiment. Note that G58 produces 2 HO2 radicals and so will have to be adjusted accordingly. Comment on the main source and sink reactions for HO2.
Note that as mechanisms get more complex (e.g. field campaign models utilising the full MCM) then the ROPA analysis can get quite large. Therefore it is often more convenient to output values in a table, which can be done using the TABLE routine (see the appropriate section of eth011001_ropa.fac where tables are set up to output the G value data for HO2 production and destruction). It is also sometimes more convenient to output G values for a longer time scale (e.g. every 30 minutes) than the concentration values.
The Rate of Production/Destruction Analysis can be performed for any species in the model (except for constrained species). Other more sophisticated techniques (such as uncertainty analysis, mechanism reduction and multivariate analysis) can be applied to the models to extract more information from the mechanism and achieve a deeper understanding of the chemistry. Examples of these techniques (applied to chamber studies) can be found in Zador et al. 2005 and 2006.