[esp-r] Re: Contamination modelling reveals multiple issues

[Geissler Achim]

Christian,
in addition to the below, are you aware of the .xml output option? I do not know to which extent the contaminant and mass and air flow variables are threaded through to this output (plant output is to some degree), but if the variables you need are, it is much quicker than using res in my opinion (if you have a "graph tool of your choice" to hand, anyway). And, as it "only" dumps all the results values, there should be no strange output depending on sequence of results analysis (which sounds scary in any case!):

Quote:
Selective extraction of data based on user-inputs prior to a simulation exist in the xml-output. It works as follows:
H3Kreports stores selected data in memory, or (optionally) on disk. Data not specified in the input.xml file are dropped. 
At the end of the simulation, H3Kreports spits out an out.xml with annual and monthly summary data, and (optionally) an out.csv file with time-step-by-time-step data. 
If the user provides a path to a valid xslt stylesheet, bps will use the libxslt library to dynamically translate into another format (e.g. .html, .svg)

A sample input.xml file follows:
 
<?xml version="1.0" encoding="UTF-8"?>
<configuration>
  <!-- format of resulting out.xml file -->
  <hierarchy>flat</hierarchy>
  
  <!--Optionally dump all data to out.xm-->
  <dump_all_data>true</dump_all_data>
  
  <!--Instructions for style-sheet transform of out.xml file -->
  <apply_style_sheet>true</apply_style_sheet>
  <link_style_sheet>false</link_style_sheet>
  <transform_destination_file>summary.csv</transform_destination_file>
  <style_sheet>/home/aferguso/esp-r/xsl/generic_summary.xsl</style_sheet>
 
  <!--Create a list of all data available for output-->
  <output_dictionary>true</output_dictionary>
 
  <!--Possibly include data from start-up?-->
  <report_startup_period_data>false</report_startup_period_data>
 
  <!--Apply timestep averaging?-->
  <time_step_averaging>false</time_step_averaging>
 
  <!--Allow regexp-like wildcard matching for list of output variables-->
  <enable_xml_wildcards>true</enable_xml_wildcards>
  <!--List of data to appear in out.xml-->
  <log_variable>building/month</log_variable>
  <log_variable>building/time/future</log_variable>
  <log_variable>building/all_zones*</log_variable>
  <log_variable>building/zone_??/flux*</log_variable>
  <log_variable>building/zone_??/air_point/temperature</log_variable>
  <log_variable>building/zone_??/casual_gains*</log_variable>
  <log_variable>climate/dry_bulb_temperature</log_variable>
  <log_variable>climate/solar/direct_normal_radiation</log_variable>
  <log_variable>total_fuel_use*</log_variable>
  <log_variable>plant/ideal*</log_variable>
  <!--List of data to appear in out.csv-->
  <step_variable>building/month</step_variable>
  <step_variable>building/time/future</step_variable>
  <step_variable>building/all_zones*</step_variable>
  <step_variable>building/zone_??/flux*</step_variable>
  <step_variable>building/zone_??/air_point/temperature</step_variable>
  <step_variable>building/zone_??/casual_gains*</step_variable>
  <step_variable>climate/dry_bulb_temperature</step_variable>
  <step_variable>climate/solar/direct_normal_radiation</step_variable>
  <step_variable>total_fuel_use*</step_variable>
  <step_variable>plant/ideal*</step_variable>
</configuration>

Regards,
Achim

> -----Original Message-----
> From: esp-r-bounces at lists.strath.ac.uk [mailto:esp-r-
> bounces at lists.strath.ac.uk] On Behalf Of Aizaz Samuel
> Sent: Monday, June 16, 2008 2:10 PM
> To: esp-r at lists.strath.ac.uk
> Subject: [esp-r] Re: Contamination modelling reveals multiple issues
> 
> Christian,
> 
> In my experience the problem is not with the way ESP simulates but in
> results
> display. I have noticed that if mass flow network results are viewed in a
> random way then unrealistic values may sometimes be observed. Contaminant
> simulation is based on network mass flow and therefore contaminant results
> may also appear different from expected when not viewed in sequential
> order.
> As far as I know the bug has not been resolved. The work around is to view
> all flows, concentrations etc. in order.
> 
> Solver instability has been reported in the past for contaminant modelling,
> try reducing your timestep to few minutes.
> 
> Although it may not have been tried before flow networks, plant and
> contaminants should work together - I plan to build such a model soon.
> 
> Regards,
> Aizaz
> 
> On Friday 13 June 2008 17:55, Christian Anker Hviid wrote:
> > Dear ESP-r community
> >
> > I am puzzled byt the contamination feature. Here are my findings:
> >
> > I have a model of a building where I use a combined air flow network
> (AFN)
> > and plant system (PLANT) with a control law that acts on the air flow
> > network. I introduce a contamination model CONT with CO2. The
> contamination
> > model and the building have the same timestep /hour.
> >
> > I have included an examplar file in the results: cellular_contam_bc
> which
> > contains AFN and CONT.
> >
> > With timestep averaged output (hourly integration)
> > AFN + PLANT: Flow looks fine. Flow is controlled by the ctrl law and
> shows
> > high value in day and low at night. AFN + CONT: Flow is wrong with
> jagged
> > spikes and values beyond reality. CO2 is also jagged. AFN + PLANT +
> CONT:
> > Flow is obviously wrong with jagged spikes. CO2 levels are also jagged
> > spikes. EXAMPLAR: CO2 is fine if plotted first, flow is oscillating.
> >
> > Output saved at each timestep:
> > AFN + PLANT : Flow is not correct, shows almost constant value. Not
> > possible to plot results when a plant model is included here? AFN +
> CONT:
> > Flow is correct and CO2 is correct if I plot CO2 before flow. Plotting
> flow
> > first spoils the y-axis scale of CO2. Adjusting scale does not bring CO2
> > curve back. AFN + PLANT + CONT: Flow is not correct, almost constant
> value.
> > CO2 seems ok if I consider the flow results. EXAMPLAR: CO2 looks fine,
> Flow
> > looks nice.
> >
> > What am I doing wrong? Contamination works ok fine with AFN, but not
> with a
> > combined AFN+PLANT. Why does AFN+CONT only work when I save output at
> each
> > timestep? Anybody with experience in contamination modelling? Thank you
> for
> > your help.
> >
> >
> > Best regards
> > Christian Anker Hviid
> >
> > Civ. ing., Erhvervsph.d.-studerende
> > DTU BYG / Alectia A/S
> > Brovej, bygning 118, lok. 213
> > 2800 Kgs. Lyngby
> > E-mail: cah at byg.dtu.dk
> > Tlf. DTU: 45 251 886
> > Tlf. Alectia: 88 191 365
> > Mob. (privat): 40 879 749
> 
> --
> 
> _______________________________________________
> esp-r mailing list
> esp-r at lists.strath.ac.uk
> http://lists.strath.ac.uk/mailman/listinfo/esp-r