I have, through a confluence of mellifluous events, acquired an MF6.2 model of a tyre I am interested in understanding on our vehicle. Hurray!
However, I expected the default interface built into ADAMS to be able to run the tyre - but it baulks and says it "can't find 'TNO_DelftTyre_Adams_interface::TYR815'". Boo!
The MF interface is clearly present and correct because it will run MF6.0 tyres I also have to hand. Hurray!
I'm obviously missing something but it's unclear to me just what...
I thought I only needed a dedicated licence for the more complex models like SWIFT, not for the quasi static magic formula.
It's looking from the MSC documentation like I need a separate dll ("app") for anything after MF6.0. I guess I missed that change, whenever it happened...
So according to this, the basic MF-Tyre support is freeware, and everything from version 6.2 on (all 8 subsequent versions) support MF6.2.
Having downloaded mftyre_mfswift-installer-2021.1-win64.exe and the accompanying manual (after contacting Siemens), run the installer, copying the DLL into C:\Program Files\MSC.Software\Adams\2021_0_1_784690\win64, editing the tyre file to refer to the current name of the DLL ("Simcenter_Tire_ADAMS_Interface") and setting USE_MODE to 14 (turning off the turnslip function) then it all works just fine.
Now to put it in a tyre test rig and see if it replicates the measured data somewhat...
To be fair, the last time I had this much messing around was something like V10 in 2004(?). It's been working pretty well since - which is why I had mostly forgotten how it works. So I am presuming I won't have this much faff going forward. Serves me right for having an MF6.2 model fitted to the test data without having tested the workflow! ;-)
In this case because we also have MF6.2 from TASS/Siemens and ran into the same problems ....
The problem is that you need to additionally install some dll/so files that you have to download from Siemens.
If you have a Windows local installation, they offer an installer that should do the job. (which you did)
In our case we have Linux & a Server/Client thing for Win10 plus no admin privileges. That's why we got their installation as a zipped file and then I linked / copied the necessary files into our MDI_USER_PLUGIN_DIR.
Search for things like
mfswift_tire_interface.dll
Simcenter_Tire_Adams_Interface.dll
TNO_DelftTyre_Adams_interface.dll
tno_delfttyre.dll
tno_userroad.dll
On Linux it's
libmfswift_tire_interface.so
libSimcenter_Tire_Adams_Interface.so
libTNO_DelftTyre_Adams_interface.so
libTNO_DelftTyre.so
If you don't have set MDI_USER_PLUGIN_DIR you can also copy those to <topdir>/win64
(Which is what you did)
BTW: As far as I remember the TNO_DelftTyre_Adams_interface is the old & outdated version.
Changing the tir-file to Simcenter_Tire_Adams_Interface::TYR815 should use the latest version.
(which is what made it work finally)
And if there's more questions about MF6.2, how about contacting their support ? ( tire.support.sisw@siemens.com )
Thanks for the reply, Martin, your wisdom is always valuable.
We've also engaged with Siemens (it's how we got the installer in the first place) and are up and running now.
I think as much as anything I was wrongfooted that a base MF routine wasn't included by default. In talking with Siemens I have understood how after TASS "took back control" by authoring definitive plugins and documentation, they were purchased by Siemens and have become much more determined than the original "open source" approach that gave rise to the wild west I was used to. That all this happened years ago is some kind of indictment of me not keeping up. I guess last time I looked at it all properly was in 2013 for the book notes.
I am poking into the corners of how to export tyre states for our colleagues co-simulating in the controls group - I may start another thread on that as it's becoming interesting...
This is an easy one to answer. I'm trying to keep us to a "unified vehicle model" on the grounds that if one person has the correct yaw inertia, everybody should have it. However to reduce my workload I don't want parametric and literal models on the go.
The run-times for co-simulation are much less distressing than they used to be so this is much more thinkable than it was when I railed against it in 2004.
They want to compare some internal state estimators with the "real" values from the tyre to improve the estimator performance, which seems reasonable to me. I wouldn't wish the A/View postprocessor on anyone that I liked, so I don't mind.