Unit Handling about fmi-standard HOT 8 CLOSED

Comment by otter on 7 Apr 2011 08:16 UTC
Since the SI Base units are a subset of the proposed {SI Base units, "rad" and "user defined units} whatever can be described by SI base units is included in the current proposal.

The issues are the following:

• "rad" must be part of the derivedUnit description since otherwise units such as rad/s or Hz (both 1/s in SI base units) cannot be distinguished (and cannot be converted). For conversion of units this is a must. For dimensional analysis (for which the SI base units have been developed) "rad" must be replaced by "1". So, "rad" must be added to the SI base units.

• There are other units occurring in applications, that cannot be expressed with SI base units, such as: "%", "ppm", "pu", "deg". From a dimensional point of view, all of them have unit "1", but this does not help when units are compared or should be converted. Therefore, in the last meeting the essential idea was to keep this flexible and add user defined base units (that cannot be expressed with SI base units). There is then, of course, the problem that there is no standardized units which potentially might give problems when utilizing the FMU in different tools. However, if it is not supported at all, then it is even worse, because then it never works!

Suggestion: Improve the description in the current proposal to make this clearer and add the most important other "base units" as normativ description (%, ppm, pa, deg).

from fmi-standard.

Comment by HansOlsson on 8 Apr 2011 10:32 UTC
Just a minor note for at least one of these: "per unit(pu)" in fact has scaling depending on which variable it is, so that two identical "pu" values does not imply that the variables have the same value in SI-units.

I thought scaling was previously done per variable, and in that case "pu" can be used directly - but for a common definitions of units one would had to defined "pu" including what and which reference, e.g. "pu - Voltage, 400V nominal".

from fmi-standard.

Comment by otter on 13 Apr 2011 07:51 UTC
Decision at FMI-Meeting in April 2011:

Only use SI base units + define them as xml attributes (not in a string).

Open issue:

How to handle "rad/s" (angle/s) and "1/s" (revolutions/s)

from fmi-standard.

Comment by otter on 17 Apr 2011 21:04 UTC
The current unit definition with 7 SI base units has the following problem (as previously discussed):

How to map the following units to the Unit definition:

• /s: mapped to 1/s with the understanding that this is "rad/s" or "revolutions/s" ???
• /s: mapped to 1/s with the understanding that this is "rad/s" or "revolutions/s" ???
• rev/min: is the factor selected so that this is expressed as "rad/s" or "revolutions/s" ???

Since rotations are a very important application, it is clear that the current status cannot be accepted. Here is a proposal along the lines from the FMI meeting in February:

1. The SI supplementary units ("rad", "sr") are added to the 7 SI base units.

2. Angle and revolutions are always expressed in "rad". This means that
   "rad/s" is "rad=1, s=-1, factor=1".
   "rev/min" is "rad=1, s=-1, factor=2*pi/60 (expressed in "rad/s" with a corresponding factor)

3. For dimensional analysis the following strategy can be used:

   1. Treating "rad" and "sr" as base units. If a dimension error
      is triggered, goto 2.
   2. Treating "rad" and "sr" as "1".

The modeller can then choose how to write his equations. For example the torque balance can be written as:

J*der(w) = t*rad_const   // rad_const is constant 1.0 with unit "rad"

Then dimensional analysis works. If someone finds this odd, the "rad_const" can be removed and the rule (2) above still accepts it.

from fmi-standard.

Comment by HansOlsson on 18 Apr 2011 07:41 UTC
There are no supplementary SI-units since 1996; the radian and steradian are dimensionless derived units.

See e.g. http://www.bipm.org/utils/common/pdf/si_brochure_8_en.pdf (page 48/88).

The importance for us is these two are not special compared to other derived units (note that some derived unit were introduced for human safety like Bequerel); and to determine why SI decided to go from supplementary units to this.

from fmi-standard.

Comment by otter on 21 Sep 2011 07:00 UTC
A proposal has been added to the FMI 2.0 draft. Basic ideas:

• UnitDefinitions contains "Unit", and optional "BaseUnit" and "DisplayUnit".

• "Unit" is a string that defines the unit of a variable.

• The optional "BaseUnit" definition contains the exponents of the 7 SI base units and the derived unit "rad", as well as factor and offset. This defines how to determine "Unit" from "BaseUnit". If "Unit" contains an "Angle" or "Revolution" quantity, then both are expressed in "BaseUnit" by "rad" (as "Angle").

• Units that cannot be represented by BaseUnit (like "p.u.", or "!Euro/Personyear") do not have a BaseUnit. This means that no operations can be carried out on these units (e.g. conversion between units is not possible).

• If variables have a BaseUnit definition, various operations can be carried out. Depending on the operation, "rad" is used as "rad" or as "1" (e.g. for connection check, "rad" is utilized; for dimensional analysis, "rad" is treated as "1", i.e., as dimensionless)

from fmi-standard.

Modified by dietmarw on 19 Mar 2012 10:11 UTC

from fmi-standard.

Modified by dietmarw on 7 Sep 2012 10:18 UTC

from fmi-standard.