View Issue Details
| ID | Project | Category | View Status | Date Submitted | Last Update |
|---|---|---|---|---|---|
| 0002510 | OpenFOAM | Bug | public | 2017-03-21 12:39 | 2017-05-03 18:46 |
| Reporter | tniemi | Assigned To | henry | ||
| Priority | low | Severity | tweak | Reproducibility | always |
| Status | resolved | Resolution | fixed | ||
| Product Version | dev | ||||
| Summary | 0002510: Subtle bug in fieldAverage function object | ||||
| Description | The recent changes to fieldAverage (https://bugs.openfoam.org/view.php?id=2459) introduce a subtle bug, where totalTime_[fieldi] is not properly initialized for fields that are not present in the Properties-dict during restart. Possible fixes: 1. In line 245 give a default value obr_.time().deltaTValue() to totalTime_[fieldi]. This corresponds to earlier behavior, but may not be accurate for variable time step cases. 2. Give an invalid value for totalTime_[fieldi], eg. negative, and detect these in initialize(). So in addition to checking for size, loop over totalTimes and initialize if necessary. 3. Disallow adding new fields to the same fieldAverage-function. Throw a warning or crash if props are not found for all fields. | ||||
| Steps To Reproduce | Steps to reproduce 1. Start averaging, run until write and kill solver 2. Add new fields to the same fieldAverage-object 3. Restart averaging, totalTime for new fields is now arbitrary leading to incorrect results or crash | ||||
| Tags | No tags attached. | ||||
|
|
Option 1 is not good because it recreates the previous error is a slightly different form. Option 2 is a bit clunky but workable. Option 3 is the cleanest but least flexible. Which option do you prefer? |
|
|
I think option 2 would be best, because the properties-dict supports having different averaging times for different fields and I see no reason to prevent user from adding new fields. Also this has worked before and the changed behavior may confuse users. |
|
|
Can you provide a patch for option 2? |
|
|
Ok, I have attached a very straightforward patch which uses -1.0 as a dummy value. |
|
|
Thanks Timo Resolved by commit c95bf5e31d627456713266e29f5ca32e1bad9216 |
|
|
I just noticed that this bug also affects 4.x. The fix should probably be also added to 4.x? |
|
|
Resolved in OpenFOAM-4.x by commit f10cfe3fb0ab9360a9b47bd0cfba4c7dfc0c56d3 |
| Date Modified | Username | Field | Change |
|---|---|---|---|
| 2017-03-21 12:39 | tniemi | New Issue | |
| 2017-03-21 12:53 | henry | Note Added: 0007956 | |
| 2017-03-21 12:59 | tniemi | Note Added: 0007957 | |
| 2017-03-21 13:03 | henry | Note Added: 0007958 | |
| 2017-03-21 13:30 | tniemi | File Added: fieldAverage.C | |
| 2017-03-21 13:30 | tniemi | Note Added: 0007959 | |
| 2017-03-22 18:39 | henry | Assigned To | => henry |
| 2017-03-22 18:39 | henry | Status | new => resolved |
| 2017-03-22 18:39 | henry | Resolution | open => fixed |
| 2017-03-22 18:39 | henry | Fixed in Version | => dev |
| 2017-03-22 18:39 | henry | Note Added: 0007968 | |
| 2017-04-20 07:59 | tniemi | Status | resolved => feedback |
| 2017-04-20 07:59 | tniemi | Resolution | fixed => reopened |
| 2017-04-20 07:59 | tniemi | Note Added: 0008033 | |
| 2017-05-03 18:46 | henry | Status | feedback => resolved |
| 2017-05-03 18:46 | henry | Resolution | reopened => fixed |
| 2017-05-03 18:46 | henry | Fixed in Version | dev => 4.x |
| 2017-05-03 18:46 | henry | Note Added: 0008076 |
