2017-06-25 06:14 BST

View Issue Details Jump to Notes ]
IDProjectCategoryView StatusLast Update
0002375OpenFOAMFeaturepublic2017-01-11 09:26
ReporterShor-ty 
Assigned Tohenry 
PrioritynormalSeverityminorReproducibilityalways
StatusclosedResolutionsuspended 
PlatformLinuxOSUbuntuOS Version16.04
Product Version4.x 
Target VersionFixed in Version 
Summary0002375: Add the record of t=0 to the probe files
DescriptionHey everybody,

it would be nice to have, if it is possible to add the first time folder (normally the zero one) to the output of the probe files? Why? If I record the temperatures at several probe positions and plot the result, the first value should be the starting value instead of the first saved one. It should be not a big deal because the function objects are called within runTime.run().

Kind regards,
Tobias
TagsfunctionObject
Attached Files

-Relationships
+Relationships

-Notes

~0007423

henry (manager)

Can you provide a patch to add this functionality?

~0007429

Shor-ty (reporter)

Yes I can, I will check it out the next days.

~0007522

Shor-ty (reporter)

Dear Henry,

I added a patch of the two modified files in order to provide the data for t = 0. I checked it with the probes:

 Probe 0 (0 0.01 0)
# Probe 0
# Time
            0 273
         0.01 285.293
        0.015 294.616


Files that are modified:

timeControlFunctionObject.C
timeControlFunctionObject.H


Cheers.

~0007524

henry (manager)

In your proposed change the start-time of the run is hard-coded to be '0'; what should happen for cases started at different times? e.g. the engine tutorial case which starts at -180 deg?

~0007525

Shor-ty (reporter)

Ah okay, never had something like that. In that case I will provide a new patch in order to make it user-defined (lookup). Thanks for the hint.

~0007526

henry (manager)

I think it would be better to use functionObject "start()" virtual function which is called on start-up rather than reproduce this functionality specifically in this class.

~0007527

Shor-ty (reporter)

Here is the new patch. In order to avoid the problem that you mentioned, we can now read the value "timeInit" to ensure that e.g. the initial condition of -180 is saved. The default value is set to 0.

~0007528

Shor-ty (reporter)

Sorry ... the uploaded patch is not working ...

~0007529

Shor-ty (reporter)

Here it is. In addition some test starting with time -3.

# Probe 0 (0 0.01 0)
# Probe 0
# Time
           -3 273
        -2.99 285.293
        -2.98 304.856

~0007530

Shor-ty (reporter)

Sorry Henry I did not notice your hint. Yes it would be better to put it into the start() function. I will do it and upload the patch. Thanks.

~0007532

Shor-ty (reporter)

Good morning,

do you mean the start() function of the functionObjectList class which calls the read() function of the class? I think that we could put the functionality into the read() function of functionObjectList. Any other suggestions?

~0007533

henry (manager)

Sorry for the confusion, I replaced the need for the functionObject::start() with start-up construction so anything you need to do at start-up of the functionObject should be done in the constructor rather than in a "start()" function. What we need is a test in the constructor for the time-step being the first of the run. I will put this in shortly.

~0007534

Shor-ty (reporter)

Ah okay, well in that case you can just move my function to the constructor of the class (the corrected tar.gz). Thanks for adding that to FOAM.

~0007535

henry (manager)

It is not clear why you are creating special control structures to write the initial value, also it is not clear why you use

    foPtr_->write();

rather than call the timeControl::write() function which includes the rest of the control logic:

bool Foam::functionObjects::timeControl::write()
{
    if (active() && (postProcess || writeControl_.execute()))
    {
        foPtr_->write();
    }

    return true;
}

which would need to be included in timeControl::saveInitConditions() otherwise.

I suggest simply adding a call to write() in the constructor:

Foam::functionObjects::timeControl::timeControl
(
    const word& name,
    const Time& t,
    const dictionary& dict
)
:
    functionObject(name),
    time_(t),
    dict_(dict),
    timeStart_(-VGREAT),
    timeEnd_(VGREAT),
    nStepsToStartTimeChange_
    (
        dict.lookupOrDefault("nStepsToStartTimeChange", 3)
    ),
    executeControl_(t, dict, "execute"),
    writeControl_(t, dict, "write"),
    foPtr_(functionObject::New(name, t, dict_))
{
    readControls();
    write();
}

~0007536

Shor-ty (reporter)

Okay.

simply adding the write() function to the constructor will not work because the writeControl_.execute() is false at the beginning and hence we will not call the foPtr_->write(). In order to prevent analyzing if the time is equal to the initial time (that we set), I made the new "saveInitCondition" function to ensure that this function is only called once during a simulation.

However, calling Ptr_.write() is related to the fact that write() simply will not be executed at the initial step. If we would use the write() function in the constructor, as you suggested, then we have to manipulate the if statement in the write() function to ensure that the Ptr_->write() is called at the initial time. In order to provide everything the if statement should then look somehow like:


bool Foam::functionObjects::timeControl::write()
{
    if (active() && (postProcess || writeControl_.execute() || (timeInit_ == time_.value()))
    {
        foPtr_->write();
    }

    return true;
}

But this will introduce more operations that are not really needed. Catching timeInit_, getting time_.value() and compare both. But it is your decision how to implement it.

I hope I explained it well :)

~0007537

henry (manager)

I tested "simply adding the write() function to the constructor" and it worked fine for me, could you provide details of the setup you are using and why this doesn't work for you? If there are cases where this does not work more control should be added to the writeControl class to allow the write at startup.

Adding "timeInit_" should not be necessary because "Time" already holds an time-step counter which can be used to test for start and restart and testing for exact equality of real numbers not reliable.

Also it is not clear if "execute()" should be called before "write()": what happens if the "write()" functions writes data generated by "execute()"?

~0007604

henry (manager)

I have tested your proposed change to functionObjects::timeControl with the simpler form of 'write' in the constructor for a range of cases and functionObjects and as expected it works for some but fails for others depending on the availability of the data to write during construction. Adding 'execute()' before 'write()' in the constructor resolves the problem for some but fails for others depending on the validity of the execution at time zero.

~0007620

henry (manager)

Waiting for feedback from reporter.
+Notes

-Issue History
Date Modified Username Field Change
2016-12-06 15:33 Shor-ty New Issue
2016-12-06 16:18 henry Note Added: 0007423
2016-12-07 19:18 Shor-ty Note Added: 0007429
2016-12-19 12:37 Shor-ty File Added: timeControlFunctionObject.tar.gz
2016-12-19 12:37 Shor-ty Note Added: 0007522
2016-12-19 13:20 Shor-ty Tag Attached: functionObject
2016-12-19 15:14 henry Note Added: 0007524
2016-12-19 15:22 Shor-ty Note Added: 0007525
2016-12-19 15:34 henry Note Added: 0007526
2016-12-19 15:36 Shor-ty File Added: timeControlFunctionObjectUserDefined.tar.gz
2016-12-19 15:36 Shor-ty Note Added: 0007527
2016-12-19 15:38 Shor-ty Note Added: 0007528
2016-12-19 15:39 Shor-ty File Added: timeControlFunctionObjectUserDefined_corrected.tar.gz
2016-12-19 15:39 Shor-ty Note Added: 0007529
2016-12-19 15:42 Shor-ty Note Added: 0007530
2016-12-20 08:18 Shor-ty Note Added: 0007532
2016-12-20 08:56 henry Note Added: 0007533
2016-12-20 10:34 Shor-ty Note Added: 0007534
2016-12-20 11:58 henry Note Added: 0007535
2016-12-20 13:17 Shor-ty Note Added: 0007536
2016-12-20 13:26 henry Note Added: 0007537
2017-01-07 10:06 henry Note Added: 0007604
2017-01-11 09:26 henry Assigned To => henry
2017-01-11 09:26 henry Status new => closed
2017-01-11 09:26 henry Resolution open => suspended
2017-01-11 09:26 henry Note Added: 0007620
+Issue History