Insert and update

The three workhorse tags write parameter values into a layer:

Tag	Writes	Selects items by	Reads existing values	Speed
~TFM_DINS	absolute values	exact names only	no	fastest
~TFM_INS	absolute values	rules: wildcards, lists, sets, topology	only for conditions	medium
~TFM_UPD	new values computed from existing ones, or absolute values where a value already exists	rules, like INS	yes (seed values)	slowest

All three are accepted in BY_Trans, SubRES_*_Trans, SysSettings, trade, regular, NSV and generated parametric files. Each has layouts that put years, attributes or timeslices across the columns (Layouts). To copy existing values to other indexes, see ~TFM_MIG.

Choose the simplest tag that can express the data: DINS when every name is known, INS when a rule picks the items, UPD only when the new value depends on the existing one.

Common columns

Column (aliases)	Meaning
Attribute	The attribute (in the plain layouts).
pset_pn, pset_pd, pset_set, pset_ci, pset_co	Process filters (filters); DINS takes only pset_pn.
cset_cn, cset_cd, cset_set	Commodity filters; DINS takes only cset_cn.
Region	Region(s) or region groups, comma list allowed.
Year (yr)	Year; 0 for the interpolation option; a range for period attributes.
TimeSlice (ts), LimType (bd), Curr (currency)	The other indexes.
Other_Indexes (oth_ind)	Extra index(es) some attributes need; several, separated by ~ (below).
Stage, SOW	Stochastic indexes.
Value (allregions, valfield)	The value for every region of the row that has no column of its own.
region names	One column per region: the value for that region.
attrib_cond, val_cond, avc_year, avc_timeslice, avc_limtype	Conditions on existing values (INS, UPD; conditions).
SourceScen	UPD only: the layer(s) to read existing values from.
Top_Check (top_chk)	INS, UPD: when filled, an item with a process and a commodity is kept only if they are connected: a (either direction), i (input), o (output); n turns the check off.
TS_Filter	DINS, INS: whole-series replacement between layers (below).

The complete list, with all aliases, is in the tag catalog.

~TFM_DINS

Direct insert: the rows are fully enumerated. Names are read literally — pset_pn holds one process name, cset_cn one commodity name — with no wildcards and no other filter columns. Regions may still be a comma list or a region group.

Because no rule has to be resolved, DINS is the fastest way to bring large volumes of data into a model, and it avoids the accidental matches that a literal _ causes in INS.

~TFM_DINS
Attribute	pset_pn	cset_cn	Year	REG1	REG2
ACT_BND	MINCOA1		2030	120	80
FLO_SHAR	EGASCC	GAS	2030	0.4	0.5

A value for a process and commodity that are not connected in the topology is dropped.

~TFM_INS

Insert: absolute values for every item a rule selects. All filters, comma lists and wildcards are available.

In this example from DemoS_001, an INS table declares three constants by row (G_DYEAR, Discount, YRFR).

~TFM_INS
Attribute	pset_set	pset_ci	Year	AllRegions
NCAP_BND	ELE	COA	2030	0
PRC_AOFF	ELE	NUC	BOH-2027	1

The first row forbids new coal power capacity in 2030 in all regions; the second makes nuclear plants unavailable until 2027. A row with an empty filter column is not restricted by that column, but at least one process filter is needed for attributes that have a process.

Several extra indexes

Some attributes have more indexes than there are columns, for example a second region or commodity. Give them in Other_Indexes, separated by ~; Veda gives each part to the index it belongs to.

Attribute	Region	pset_pn	cset_cn	Other_Indexes	Value
IRE_FLOSUM	REG1	IMPCOA1	COA	imp~out~INDCO2	0.1
IRE_CCVT	REG2		GAS	REG1~GAS	1

~TFM_INS-TXT

INS for the attributes whose values are text, which override what the declaration tables set: PRC_PCG, PRC_TSL, PRC_VINT, COM_LIM, COM_TSL, COM_TYPE. Accepted in SysSettings only.

~TFM_UPD

Update: compute new values from existing ones. A UPD row selects items exactly like INS, then looks for existing values of its attribute for those items, and writes a new value for each one found. Where no value exists, nothing is written.

The example sets prices (ACTCOST) for the dummy import processes (IMP*Z) wherever they already have one.

Value cells in ~TFM_UPD

The value cell may start with an operator:

Cell	New value
*1.2	existing × 1.2
/2	existing ÷ 2
+5	existing + 5
-5	existing − 5
~-5	−5 (the ~ marks an absolute value)
40	40

A leading minus is subtraction

In UPD (and MIG) cells, -5 means "subtract 5". To write the value −5, use ~-5.

Where existing values come from

From the seed values of the layer: BASE, the repositories, trade and demand layers, and regular layers whose name sorts at or before the UPD's own layer. When several of those layers hold a value for one item, a regular layer beats a trade, demand or repository layer, which beats BASE; among regular layers, the one whose name sorts last is used. SourceScen (a layer name or comma list) restricts the source to the named layers; use it whenever the source matters.

A UPD row does not see values written by other UPD rows of the same table, and runs side by side with ~TFM_MIG (processing order).

Updating is not the same as replacing

Any of these tags replaces a value only in the sense of the case: a Scen_HighGas row writes a value in layer HighGas, and a case that includes HighGas after BASE uses it. UPD is needed only when the new value must be computed from the old one, or written only where an old one exists. In BY_Trans and SubRES_*_Trans, which belong to the same layer as their parent, an INS or DINS row for the same item overwrites the template's value directly.

Performance

UPD is the heaviest tag: each row fetches existing values from a potentially large table. A UPD table with few rows, each touching many records, is fine; a long table with one row per year and sub-sector can take minutes. When the update is really a fixed set of new numbers, write it with DINS or INS instead. UPD tables shows a worked example: an update that took four minutes, replaced by a ~TFM_FILL-R + ~TFM_DINS pair that takes a fraction of the time.

Layouts: -AT, -TS, -TSL

The plain tags have one value per row, under region columns or AllRegions. The layouts put another index across the columns instead, so source data can be pasted as it is:

Suffix	Column headers are	Exists for
-AT	attribute names	~TFM_DINS-AT, ~TFM_INS-AT, ~TFM_UPD-AT, ~TFM_MIG-AT
-TS	years (a time series), plus 0 for the interpolation option	~TFM_DINS-TS, ~TFM_INS-TS, ~TFM_UPD-TS, ~TFM_MIG-TS
-TSL	timeslices	~TFM_DINS-TSL, ~TFM_INS-TSL

The naming is counter-intuitive: -TS means years ("time series") and -TSL means timeslices.

With a layout, the index that is across the columns is not a column of its own; the region then comes from a Region column or the tag text (~TFM_INS-TS: Region=Reg_Dev). The attribute of a -TS or -TSL table can be on the tag (~TFM_INS-TS: attribute=SHAPE) or in an Attribute column, which lets one table carry several attributes.

~TFM_INS-TS — years across:

~TFM_INS-TS
pset_ci	pset_co	Attribute	cset_cn	2020	2030	0	LimType
GASNGA	HH2	FLO_EMIS	CO2CAP	0.9	0.95
ELCCurt	HH2	FLO_SHAR	ELC	0		3	LO

Use -TS for any series that should be interpolated

In a -TS table the year columns form a series, and column 0 is the rule for interpolating and extrapolating it. In a plain table with year columns, each column is an isolated data point. For a bound, the years between the points are then simply unbounded.

~TFM_DINS-TS — the same, fully enumerated, here with the attribute and limit type on the tag:

~TFM_DINS-TS: attribute=COM_BNDPRD; LimType=LO
Commodity	Region	2019	2020	2030	0
pc_2eh	Asia_Dev	0.374	0.293	0.311	5
pc_2eh	Belgium	0.041	0.037	0.054	5

~TFM_INS-AT — attributes across, several attributes per row:

~TFM_INS-AT
pset_set	Region	Year	PRC_AOFF	START
E_Coal	Bangladesh,Indonesia,Vietnam	2040-EOH	1
E_Coal	Bangladesh,Indonesia,Vietnam			2025

The first row switches coal plants off from 2040 to the end of the horizon; the second sets their first availability year.

TS_Filter: replace a whole time series

Normally, when a later layer in a case gives a value for an item, it replaces the earlier layer's value year by year: a later 2030 value leaves the earlier 2020 and 2050 values in place, and TIMES interpolates across the mix. The TS_Filter column of ~TFM_INS and ~TFM_DINS changes this for the rows that carry it, for each item (all indexes except year):

TS_Filter	Effect when the case is assembled
1	The series comes only from the latest layer in the case whose rows for that item have TS_Filter 1; rows for the item in every other layer are dropped.
0	Rows for the item in layers earlier in the case are dropped; later layers still override year by year.
empty	Normal behaviour.

Year-0 rows (interpolation options) are never dropped.