DIGITAL.CSIC. Repositorio Institucional del CSIC
Datos de investigación
Set de datos (Dataset)
Supplementary information about The PIONEER database
Data processing: Computed columns: By parsing the input data it is possible to increase the amount of columns in the database without changing the source format. The following columns are computed (or updated) in the back-end when the database is loaded from disk to memory.-- Assumptions for calculations:
A number of parameters are (re)calculated based on provided metadata for publications: the residence time (1), the SEI in J L-1 (2), energy efficiency of CO2 conv. % (5) but also the frequency range column (section 8.1).
Each of these calculations rely on data that potentially is given as a range of values, see section 1. By convention, the mean of the range is taken for each of the parameters. All entries for which a (re)calculation has been performed on either the x or y data, have the Calculated field set to ‘Calculated’ (rather than ‘Original’).
Filtering the database on only Original data, this calculated data can be replaced with calculations by the
user based on other aggregation, if desired.
As mentioned for equation (2), the calculation that converts power (in W) to SEI (J L-1) simply relies on
the gas flow rate Φ (in sccm or mLmin-1) and therefore does not take into account if Φ is defined w.r.t. standard
conditions or the actual conditions in the discharge regarding pressure and temperature. Since generally in this
case both the power and gas flux are know, the SEI can be calculated (except for batch reactors where there is
no flow), which allows for a comparison between a larger body of experiments.
Contrary to this, the calculations for the residence time (1) and energy efficiency of CO2 conversion both
take into account the discharge conditions. This discrepancy is a conscious one: the SEI (in J L-1) is often
reported as a macroscopic process parameter that is calculated from the power and fluxes put into the reactor.
For τ res taking into account the conditions inside the plasma is paramount, since it can provide the information
on how long a particle is exposed to plasma conditions. Do note that in lieu of the plasma –or active zone–
volume, for some experiments an approximation such as the reactor volume is used, see column relevant
volume in section 4.
Likewise for the efficiency, it is important to account for the actual measurement conditions to establish the
number of particles into which energy has been channelled. For in situ versus effluent conditions this can be very different. Overestimating the number density favourably improves efficiency, whilst an underestimation similarly negatively impacts it. While the way in which the PDB is structured and reports metadata is not without caveats (section 4.3 of the paper), the distinction between in- versus post-plasma dissociation measurements can be partially addressed by tailored filtering and aggregation of the data contained within.-- Normalisations:
Several normalisation functions are available in the interface of the database. As described in the main text, the aim to provide a tool for easy calculation of normalised data within the same interface and compare with different data sets. For flexibility sake, a wide set of normalisations are provided, without restriction on whether they are sensible in a given context. The purpose of each normalised function is described briefly and summarised in table 1 along with the equations utilised.-- Under a Creative Commons license CC BY 4.0., The Pioneer database (PDB) is divided into two parts: performance data and metadata. The performance data originate from measurements reported in literature, where they are typically provided in form of plots or
tables. The PDB and its dedicated online tool allow to compare large amounts of performance data to derive
trends leading eventually to process optimisation. Performance data is provided in form of plain text files with
only two comma-separated columns of numbers with a point (.) as decimal separator, without a header (as it
is inferred from the metadata). The first column contains what is henceforth called process parameters. These
are the independent variables of the experiments, i.e. the x-values like power, pressure etc., see section 7.
The second column contains the so-called performance parameters. These are the dependent variables of the
experiment, i.e. the y-values like conversion, selectivity etc., see section 7. The metadata contain additional
information about the measurements that are crucial for their interpretation. Metadata are provided in table
format following the template discussed subsequently.
Before elaborating on the actual data input, the structure of the PDB metadata is discussed. The metadata is
grouped thematically in categories. Within each category, information is entered into fields, i.e. the columns
of the table. Essential and conditional fields are distinguished. Essential fields contain crucial information for
the assessment of the plasma-catalytic process. In the best case scenario, all of them are given in the respective publication. When fields are listed in the description of a category from section 2 onward, essential fields are indicated by a regular bullet point (•). Conditional fields are by no means less important than essential ones, but can rather be left empty depending on other fields. For instance, most fields in the catalyst category are left empty, when no catalyst is used. Thus, conditional fields are meant to save time. Listed in the following,
they are indicated by a plus (+). In conclusion, all fields are strongly recommended as data useful for valuable
comparison with other work from the community. A subgroup of essential as well as conditional fields are those
fields that contain the process parameters defined in the first paragraph of this section. Generally speaking,
process parameters are the experimental settings in the pursuit of highest performance. The user of the PDB
thus encounters process parameters on two occasions: on the one hand as typical x-values in the performance
data and on the other hand as input to fields of the metadata. Hereafter, fields that contain process parameters
and parameters are used synonymously. The total of fields belonging to the same measurement make up what
is hereafter called a data set, corresponding to a row of the table. Note that here the input of data is addressed.
In the back end, a data set is broken up into (x,y)-pairs for more flexibility in data handling. Data is exported
also in that format.
To ensure comparability, a template is used for inputting information into the PDB metadata. With respect
to information entered, fields can be divided in numerical and textual fields. They are filled with numbers or
text, respectively. For example, parameters are usually numerical fields. A numerical field contains either (i)
one number x if the numerical value is known and does not change during the experiment; (ii) a range of values between xmin and xmax –given as array-like notation rxmin; xmaxs– if the numerical value is not exactly known, or when it changes in the course of the measurement; or (iii) NA if the numerical value is not known. For further use, an aggregate function is applied to array-like data to obtain a single number, by convention the mean.
A textual field contains a string of text. There are a few instances where text can be entered freely as long
as some format is followed. However, usually the field is filled by selecting from a pre-defined list of options
in the template. Most of these lists are fixed but some might be extended in the future depending on the
experimental data provided. This paragraph just gives a general overview. In the in-depth discussion of the
fields of each category, it gets more clear what exactly is supposed to be filled in each field. The metadata of the PDB are divided into six categories
• data identification
• gas mixture
• plasma source
• catalyst
• separation unit
• output data, 1. General
2. Data Identification
3. Gas Mixture
4. Plasma Source
5. Catalyst
5.1 Catalyst Coupling .
5.2 Catalyst Composition
5.3 Catalyst Pre-treatment Before Reaction
5.4 Catalyst Conditions
5.5 Catalyst Characterization
6. Separation Unit
7. Output Data
8. Data processing
8.1 Computed columns
8.2 Assumptions for calculations
8.3 Normalisations, When data is reported according to the specified scheme above, the combined data and metadata can be read
from disk and processed with some scripting. Most notably this performs data type coercion and extraction
from a more flexible ‘human-readable’ format to a consistent, ‘machine-usable’ scheme. Some of the computed columns are redundant to some extent –the authyear column for instance is just a concatenation of the first author name and publication year (yyyy) columns– but these are provided for ease of filtering or grouping data, avoiding frequent (re)computation., This project has received funding from the European Unions Horizon 2020 research and
innovation programme under the Marie Skodowska-Curie grant agreement No. 813393, Peer reviewed
Proyecto:
EC / H2020 813393