How a table is read¶

All model data is written in tables: rectangular blocks of cells on a worksheet, each introduced by a tag. This page says how Veda finds a table, which cells belong to it, how the tag text and the column headers are read, and which rows and columns are skipped. What the values mean is covered in Values, wildcards and filters.

The tag cell¶

A tag cell is a cell whose value, after trimming spaces, begins with ~. Formulas count by their result, so a tag can be built by formula.
The tag name is the text after ~ up to the first : (or the end of the cell). Tag names are not case-sensitive: ~fi_t, ~FI_T and ~Fi_T are the same tag.
The tag text is everything after the first :. It gives values that apply to the whole table (below).
A cell beginning with ~ whose name is not a Veda tag is ignored, with its table.
A tag that exists but is not allowed in the file type where it appears is reported in the sync log and ignored (tag catalog).
Cells are read by their displayed value. A cell holding an Excel error (#N/A, #DIV/0!, …) is read as empty.
Hidden sheets, rows and columns are read like visible ones.

Switching a table off

Because a tag must begin with ~, anything in front of it switches the table off. The house convention is the prefix Deact: Deact~TFM_INS. The cell still shows what the table would be, and a formula can switch it on and off (parametric scenarios).

Where a table starts and ends¶

The table occupies the Excel current region of the tag cell: the block of non-empty cells around it, bounded on every side by an empty row or column (or the sheet edge). In practice:

Leave at least one fully empty row and one fully empty column between neighbouring tables, including tables placed side by side. A single stray value in the gap merges two tables into one region.
If a tag cell lies inside the region of another tag, Veda stops the sync and reports both tags.
The table starts at the tag's row. Cells of the region above the tag row are not part of it.

Tags come in two kinds, which differ only in where the table starts horizontally:

Corner tags (most settings tags, ~TFM_TOPINS, ~TFM_MIG, ~TFM_AVA, ~TFM_COMGRP, ~TFM_FILL-R, ~UC_Sets, ~TradeLinks, …): the table starts at the tag's column. Region columns to the left are not part of it.
Floating tags (~FI_T, ~FI_Process, ~FI_Comm, ~TFM_INS, ~TFM_DINS, ~TFM_UPD and their variants, ~UC_T, …): the table spans the whole width of the region, so the tag may sit above any of its columns.

The tag catalog gives the kind of every tag.

Rows¶

The header row is the row directly below the tag cell. Data rows follow it to the bottom of the region.
Single-value tags (~StartYear, ~ActivePDef, ~EndYear, ~DefaultYear) have no header: the value is the cell directly below the tag.
A table with no data rows contributes nothing.

Tag text¶

The tag text sets values for the whole table. It has two forms.

Key–value form. If the text contains =, it is a list of key=value items separated by ;. Keys are column names (any of their aliases); a value may be a comma list. Spaces around keys and values are ignored.

~TFM_INS-TS: attribute=FLO_EMIS; limtype=UP
~TFM_DINS-TS: attribute=COM_BNDPRD; LimType=LO
~TFM_INS-TS: Region=Reg_Dev,Reg_Eme
~TFM_FILL-R: w=RoadTra_BY; hcol=region

An item acts as a column of that name holding the same value in every row. A region given here may be a region group or a comma list.

Token form (~FI_T, ~UC_T). If the text has no =, it is a list of tokens separated by ~. Each token must be a known element: a currency, a limit type, a region, a timeslice, a year or an attribute. Veda places each token in the index it belongs to.

~FI_T: USD24          every value is in USD24 unless a column says otherwise
~FI_T: UP             every attribute that has a limit type defaults to UP
~FI_T: MEUR2000~FX    both
~UC_T: UC_RHSRT       the default attribute of the table

A token sets a default only for attributes that have that index: ~FI_T: UP leaves costs and efficiencies untouched because they have no limit type.

Precedence. In ~FI_T and ~UC_T, a qualifier in a value-column header beats the row's own cell in that index's column, which beats the tag text. In the other tags, the row's cell beats the tag text, and an empty cell falls back to the tag text. Region-named columns take the place of a Region column.

Always put the currency on cost tables

A cost table without a currency is read in the model's default currency (the first entry of ~Currencies, often USD00). TIMES then converts the numbers through its deflators, which can silently change costs by tens of percent. Write it on the tag (~FI_T: USD24, ~TFM_INS: curr=USD24), in a ~FI_T header (INVCOST~USD24), or in a Curr column.

Column headers¶

Known columns and aliases¶

Each tag has a set of known columns, and most have several accepted names (aliases): TechName, process and pset_pn all name the process column of ~FI_T. Headers are not case-sensitive. The tag catalog lists every column of every tag with its aliases.

Two headers in one table that name the same known column (identical, or two aliases of one column) are a fatal error: the sync stops and reports them. (The check applies to tables that have data rows.)

Value columns and header qualifiers¶

A header that is not a known column is a value column: it says what the numbers below it are.

In ~FI_T and ~UC_T, a value-column header is split at ~ into tokens, and each token must be a known element: an attribute, a year, a region, a timeslice, a limit type, a currency or a commodity.

Header	Meaning of the numbers in the column
`INVCOST`	attribute `INVCOST`
`PASTI~2015`	attribute `PASTI`, year 2015
`ACT_BND~UP~2030`	attribute `ACT_BND`, limit type `UP`, year 2030
`NCAP_COST~USD21`	attribute `NCAP_COST`, currency `USD21`
`UC_RHSRT~2030`	(in `~UC_T`) right-hand side for 2030
`REG1`	region `REG1` (one column per region)

A header whose tokens do not all resolve is reported as "column is not recognized or invalid", and the column is not read.

In the other tags (~TFM_INS, ~TFM_DINS, ~TFM_UPD, ~TFM_MIG, …), a value-column header is one element, not split at ~: a region or region group, AllRegions/Value, or — in the layouts — a year (-TS), an attribute (-AT) or a timeslice (-TSL). A header such as PASTI~2015 is not read in these tags; give the year in a Year column or use a -TS layout.

Comment columns¶

A column whose header starts with * or \I: is a comment column and is ignored. In ~FI_T, any header containing *, \I: or % is ignored. Such headers may repeat.

Columns ignored for the whole model¶

Header names listed in ~NOpCol are ignored in every table of the model, and their warnings are suppressed. A listed name is still read in a tag for which it is a known column: listing Sets silences a stray Sets column next to a ~FI_T table without affecting the Sets column of ~FI_Process.

Comment rows¶

A data row is skipped when, in one of the table's index columns, its cell starts with \I:, or starts with * in a column where * cannot be a wildcard. Precisely:

In columns that never hold patterns (attribute, year, region, limtype, timeslice, currency, other_indexes, value columns whose header contains ~, …), a cell starting with * or \I: makes the row a comment.
In columns that may hold patterns (pset_*, cset_*, commodity, value, …), only \I: does, because * there is a wildcard.

\I: is therefore the safe comment marker everywhere. Model Info → Veda Tags shows which marker applies to each column of each tag.

Values carried down the rows¶

In a few columns an empty cell takes the value of the nearest non-empty cell above it in the same table:

Tag	Column
`~FI_T`	`TechName` (process)
`~FI_Process`	`Sets`
`~FI_Comm`	`Csets`
`~UC_T`	`UC_N`

So a process is named once on its first row in ~FI_T, and the rows below add more inputs, outputs or attributes to it. No other column is carried down; in particular, years are not.

Wide and long layouts¶

An index that is usually a column can instead be spread across columns, one per element. Veda turns such a table into the equivalent long table (one row per filled cell); empty cells produce nothing.

Regions. A header that is a region name, or a region group (expanded to its regions), holds the value for that region. An AllRegions column holds the value for every region of the row that has no column of its own.
Years. In the -TS layouts, headers that are years (and 0) hold the time series.
Attributes. In the -AT layouts, headers are attribute names.
Timeslices. In the -TSL layouts, headers are timeslice names.

Sheets¶

Every worksheet is read; sheet names carry no meaning, with two exceptions: in the trade-links file the first letter of the sheet name decides whether links are two-way (B…) or one-way (Trade), and in SubRES files the sector of a declaration is taken from the sheet name.