[Bldg-sim] A critical juncture for Modelica/FMI next week

Per Sahlin per.sahlin at equa.se
Sat Mar 10 03:15:20 PST 2012


Hi Pavel,

It is difficult to explain these things in totally concrete terms. It has to do with simulation model scalability, portability, abstraction and ease-of-use.

Both Simulink and Modelica tools offer ways of describing and simulating the behavior of a wide class of dynamical systems. This allows you to tailor exactly the equations you want to use, for example in the way you outline below. This total freedom is often needed in building simulation problems, since the number of different component models needed for various studies is for all practical purposes infinite. No tool with a fixed set of models will ever be sufficient.

Now, a full-fledged building zone model like that of E+, or our tool IDA ICE, will not require 2-3 equations, but rather on the order of a thousand variables per zone. A fair number of these represent things that are meaningful to observe directly, such as the temperature of the third layer of a particular wall and others are just help variables that are needed to compute for example a sliding average. If you then want to model a thousand zone building, you get by all standards a fairly complex and large system of ordinary differential and algebraic equations (DAEs) to solve. This is where the scalability comes in; not all approaches will be able to handle this size of problem.

When you have this many equations to keep track of, no single person or even team will have the required expertise about each type of process that needs modelling. Portability and abstraction becomes needed. If you can specialize in a particular field, such as advanced heat pump modelling, and then package your models in a way  that will plug right into relevant neighboring models and in a variety of tools, many people can use them without knowing each detail. This is the ultimate purpose of Modelica and FMI.

A building simulation model must also be possible to set up and operate by a person that knows absolutely nothing about Modelica, DAEs and variable timestep solvers. Therefore, it must also be possible to package these equation-based models in a way that, on the surface, looks much the same as traditional BEM tools - the ease-of-use.

I hope this was of some help. If you are interested to learn more about these things, browse the conference proceedings available on https://www.modelica.org/publications , where you can see how Modelica has been applied in many different domains.

Best regards,

Per




From: pdybskiy at gmail.com [mailto:pdybskiy at gmail.com] On Behalf Of Pavel Dybskiy
Sent: den 9 mars 2012 12:57
To: Per Sahlin
Cc: bldg-sim at lists.onebuilding.org
Subject: Re: [Bldg-sim] A critical juncture for Modelica/FMI next week

Per,

It is impressive ... but it sounds too abstract to me to digest.
Can you help me to think about FMI in terms of a BEM (building energy modeling) task.

For example:
1) There is a  system of ODEs (2 o3 equations), which represents a simple (say 3R2C or 3R3C) thermal network (TN), which in turn models a thermal zone.

2) There is another set of equations and inequalities that specifies HVAC system.

3) There is a given weather file (temps, radiation, etc) which provides some set of time series variables (1hour sample time) for those ODEs.

4) There are 2-3-4 schedule files, which  specifies other sets of variables/parameters  ( 1hour - 1day - 1month sample time ) for those thermal zones and HVAC.

Today I am using Matlab/Simulink to simulate this (1-4) problem.

In case if I need more complex building and HVAC models I may think about BCVTB(LBNL) (people already do this) to co-simulate various combinations of E+ and Matlab models.

What role and how can FMI/FMU play in such a game?

Thank you

Pavel


On Fri, Mar 9, 2012 at 5:30 AM, Per Sahlin <per.sahlin at equa.se<mailto:per.sahlin at equa.se>> wrote:
Dear all,

There is a meeting of the FMI (www.functional-mockup-interface.org<http://www.functional-mockup-interface.org>) design group Monday and Tuesday next week where most likely the level of ambition of FMI 2.0 will be settled. Different application areas have different needs and priorities and I think it would be wise to provide some input from the building simulation community to this meeting. Let me first give some background.

The ultimate purpose of equation based simulation and Modelica (www.modelica.org<http://www.modelica.org>) is to enable model exchange between users and tools. However, the Modelica language is by necessity a very complex and rapidly evolving "beast." It is essentially impossible to keep tools in good sync with each other in their interpretation of Modelica. In practice, this means that porting complex model libraries, such as the MSL Fluid and Media libraries and derivatives, from one tool to another often involves man years of work, if possible at all. This is where FMI comes in. FMI provides a means to exchange pre-compiled model units and this can be radically simpler than to achieve a common understanding and interpretation of the underlying Modelica code. FMI 1.0 has proven this since a number of years. However, for our application FMI 1.0 is severely limited.

There are three key areas where FMI 2.0 has previously been planned to be enhanced in ways that in my opinion are essential to the building simulation community:


-          Introduction of physical connectors (now only direct variable-to-variable connections are supported)

-          Support for arrays, the size of which is determined by parameters that can be changed after compilation (only scalars in FMI 1.0)

-          Support for sparse analytical Jacobians

In the building simulation world of multiple-port tanks and zones, where many components are built around  discretized PDE:s, FMI without these features is unlikely to become very useful. At the same time the FMI design group is facing some seriously conflicting goals:


-          FMI must not be too rapidly evolving - otherwise much of the purpose is lost

-          The standard must not be too complex for vendors to support

-          There is an urgent need to upgrade v. 1.0 to something more powerful

Therefore, there is a distinct risk that some of the planned enhancements of 2.0 are postponed. For reasons of stability, it could take a very long time before the next major release.



If you share my concern about these issues, please write a few lines to underline why we need these features to me.  I will compile for the FMI design committee.



Best regards,



Per Sahlin



-----------------------------------------
Per Sahlin
EQUA Simulation AB
per.sahlin at equa.se<mailto:per.sahlin at equa.se>
+46   8 546 20 111<tel:%2B46%C2%A0%C2%A0%208%20546%2020%20111>
+46 70 422 03 02<tel:%2B46%2070%20422%2003%2002>   (mobile)
+46   8 546 20 101<tel:%2B46%C2%A0%C2%A0%208%20546%2020%20101> (fax)
plurresahlin (skype)
Råsundavägen 100
169 57 SOLNA, Sweden
http://www.equa.se
-----------------------------------------





Subject: Re: teleconference about IEA proposal "Modelica-based next-generation tools for new and existing buildings and communities"

All,

Thanks to those who contributed to today's conference call. Attached please find the meeting minutes. Please let me know any important omissions or corrections.

As discussed, I also added a google doc where each of you can add your interest in participation and in leading a particular subtask. There is also a column in which you can specify your funding situation for the respective work, and a column in which you can add demonstration objects that could be included in the revised proposal. The link to this spreadsheet is
   https://docs.google.com/spreadsheet/ccc?key=0AnSvCGCOHhzXdHBOSDRuLXJOd2tvSlBzUXBHaU93NVE
Anyone with the link can edit and view the document. So don't add sensitive information or proprietary information. You are free to share this document, as well as the proposal, at
  https://docs.google.com/a/lbl.gov/document/d/1ia1jyXrvf1jUynN_1-3QBx0quYdXlOdYgPTacpE6lLU/edit?hl=en_US&pli=1
with your ExCo delegate and/or your potential funding agency.

Depending on how many people work on the respective subtasks, some subtasks might be combined in future revision. For now I kept the structure that resulted from the Sydney meeting.

All the best,

Michael
On Thu, Feb 2, 2012 at 3:16 PM, Michael Wetter <mwetter at lbl.gov<mailto:mwetter at lbl.gov>> wrote:
All,

We would like to schedule a teleconference with those of you who are interested in participating in the IEA proposal "Modelica-based next-generation tools for new and existing buildings and communities" that was discussed in Sydney last November.

To schedule the call, please respond by Wednesday February 8 to the poll at  http://www.doodle.com/c5fvhz5uuzuq7ykr.
If you like to be removed from future emails, please respond to me.

The proposal is at https://docs.google.com/a/lbl.gov/document/d/1ia1jyXrvf1jUynN_1-3QBx0quYdXlOdYgPTacpE6lLU/edit?hl=en_US&pli=1 and attached to this email. This version addresses the feedback from the Sydney meeting, as well as feedback received from the Fraunhofer Institute in Kassel.

The items to be discussed at the phone call are below.


 1.  Outline of next steps.
Proposal due April 16, presentation to IEA ExCo meeting on June 14.
If approved, there will be a 1 year planning period (we will check whether this can be accelerated), followed by 3 years of research and 1 year of reporting.
 2.  Expression of interest of research groups to work on the proposed research.
Individual research groups should express what tasks they like to contribute to, and are likely to get funded from their respective country.
 3.  Indication of subtask leaders.
Note: Subtask leaders need not be finalized by June, but an indication of what country will lead what task is recommended. The subtask leaders can be finalized during the planning period.
Subtask leaders need a minimum commitment of 3 months/year.
 4.  Inform your ExCo delegate.
The ExCo delegates will need to verbally confirm at the June meeting that funding will be available in their country and that the proposal is in the interest of their country. For the proposal to be accepted, around five (minimum 3) countries need to commit to participate. It therefore is essential that you discuss the proposal and your interest to participate with your ExCo delegate prior to the June meeting. See http://www.ecbcs.org/contacts.htm for a list of ExCo delegates.
 5.  Other points of discussion.

Please let me know if there is anything else you would like to add to the agenda.

All the best,

Michael Wetter & Jan Hensen



_______________________________________________
Bldg-sim mailing list
http://lists.onebuilding.org/listinfo.cgi/bldg-sim-onebuilding.org
To unsubscribe from this mailing list send  a blank message to BLDG-SIM-UNSUBSCRIBE at ONEBUILDING.ORG<mailto:BLDG-SIM-UNSUBSCRIBE at ONEBUILDING.ORG>



--
Best Regards,

Pavel Dybskiy
M.A.Sc<http://M.A.Sc>. Candidate, Department of Architectural Science
Faculty of Engineering, Architecture and Science, Ryerson University
E-mail: pdybskiy at ryerson.ca<mailto:pdybskiy at ryerson.ca>;
telephone: 647-984-4940
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.onebuilding.org/pipermail/bldg-sim-onebuilding.org/attachments/20120310/c59c64e7/attachment-0002.htm>


More information about the Bldg-sim mailing list