1A Statics Notes

Apr 12th 2005

Janet Lees (jml2) is revising the IA Statics lab. Currently there are no computing requirements for this lab, however she would like to introduce a computing element. The vision -

There will be three groups of the 8 students.
There will be three unique trusses.
Each truss will have its own computer.
Each group of set of students will work on one truss at a time (until each group has worked on every truss).
Each group has its own set of weights.
Every bar on the truss has a digital gauge which measures strain.
As weights are added, so the students take a reading from each bar.
The students input the data into the workstation next to the truss.

As the data is collected, so it is sent to a fourth machine. This machine produces and displays (on screen) a graph based on the inputted data.

Eventually (although not needed for the first phase) Janet would like the results and graphs to get put on the web. This is because this lab is used for exposition, during which the students are required to write a report based on the lab.

Bob Hepburn has offered to lend Janet 4 PCs for the lab. She would like to try the lab out on the Headstart and Sutton Trust people (6th formers who are on a "try before they buy" day out around June/July time).

Today I produced a very drafty prototype - see the 1A Statics page. It's a PIN-protected facility that has tables for data entry. Tables can be made 'read-only'. Output is currently text-only but graphics would be welcomed.

Recorded with each piece of data stored is who stored the data, and there's provision for storing the person's 'group' too. The "group" concept isn't yet used - maybe 'read-only' pages could be unlocked only by those in the same group as the original locker? Navigation's not tidy either.

It uses 2 mysql (database) tables - Statics1AData and Statics1ALocking

April 14th

Using gd looks like the easiest way to provide on-the-fly graphics in various formats. It's not currently installed on the webserver, but the odds are it will be within a few weeks.

April 15th

Some other points and issues

The experiment's used heavily for the first few weeks of the first term. Use tapers off.
There needs to be provision for immediate graphical output and also for output to be later used in expositions. The data will remain in the database long after the experiment. A drop-down menu on the OUTPUT page could offer files in a variety of formats - raw data in CSV files for spreadsheets, and *.m file for matlab; graphics in whatever formats that gd offers - PNG, JPEG and GIF, I think. I've added an actionless menu to the output page just to show how things must look.
How are groups to be set-up? A supervisory mode would be useful, so that a staff member could change the group of a student if need be.
The data could be time-stamped to monitor student progress etc
Simultaneous edits/saves of the same data need to be worried about. Currently it's possible for one student to save data that's just been 'locked' by a student on another machine.

April 18th

Another database table (1AStaticsUser) added, to record when users were active and to store who the supervisors are.

April 19th

In the development version (which you're welcome to try) I've added a facility so that supervisors (currently just me!) are offered a chance to go into supervisor mode. Currently the mode lets supervisors list users by date of last access.

April 21st

Access controls in general need to be thought about - what type of access should an individual have to their own group's data and to other data? What should supervisors be able to do - move someone from one group to another? Can the computer do some admin chores?

April 25th

The Development version's been tidied up a little. Users can't save if the data's currently locked by the user. Also the menu that offers file download formats now produces files in the required formats. The CSV (spreadsheet) file might contain the correct contents. The other varieties contain dummy output.

April 28th

In the Development version supervisors now can add/remove supervisor powers for users - note that they can currently remove their own supervisor powers.

July 19th

In the Development version supervisors now can add new users. Here's a checklist of things to worry about

Administrative Issues
- Users will need to know their PINs or passwords. If they have forgotten them, they'll have to visit the operators
- The system needs to know beforehand which group each student is in. What are the groups called and how soon are they known? Are 1AStatics staff prepared to set them up given suitable support?
Supervisor Mode
- Currently any registered user can be made into a supervisor by another supervisor. What further extra powers should supervisors have? Currently they can list the last time each user logged in.
Input Mode
- Data is lost if the browser is closed or the page changed before a "Save" is done.
- Users can currently "lock" the data if they think it's complete. Who should be able to unlock it? Just the individual who locked it?
- Should there be a "clear table" option
Output Mode
- I've forgotten what kind of processing is required. What should be calculated? Should groups be able to use data provided by other groups even if they're not allowed to see it?
- What formats should be provided for output? Currently CSV (for spreadsheets), Matlab, PNG and GIF are on the menu, though only the 1st 2 option are easy for me to complete.
Cosmetics - I've not yet worried much about making the pages easy to read or use.

July 22nd

After discussion with jml2 I've changed a few things

http://www-h.eng.cam.ac.uk/help/tpl/php/1AStatics/ has been updated so that it has the features from the "development" version.
A supervisor can now get a list of the current day's loadcell calibration factors. I'm the only user today, and I've only filled in the steel phase 1 values properly, so the output is
```
   Today's users are
    tpl (group 1)

    * Phase 1 steel: loadcell calibration factor = 133.8688085676
    * Phase 1 aluminium: Data for loadcell value for load 0 is invalid
    * Phase 1 CFRP: Data for loadcell value for load 0 is "" which is
      non-numeric
```
Graphs would be better, but this is a start - the supervisor can monitor progress and compare results.
The last supervisor can no longer remove their supervisor status

August 1st

GD's installed. The GIF and PNG options on the output page now produces sample graphical output -

August 3rd

Tidying up. Work done on the output options - can now choose the material, and the matlab-format option now produces a file containing the supplied data. Here's a sample output file

% matlab file for tpl showing Steel data
kn=[0, 0.249, 0.498, 0.747, 0.996, 1.245, 1.495, 1.744, 1.993, 2.242];

displacements=[-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
0, -1, 3, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
100, -1, 4, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
200, 300, 5, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
300, -1, 6, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
400, -1, 7, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
-1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1, -1;
];

I need to make the graph provision more general, but options could easily ramify. What do users and supervisors need? Variables include

Material
Member
Phase
X and Y axis units

but the behaviour of steel member 2 and aluminium member 5 may need to be compared on the same graph (indeed, I suspect graphs will most often be used for making comparisons). As a start I'll add options to the interface on the "output" page.

August 5th, 9th

New house style added
Several graphical output options added for students. The graphs now show the data that's been typed in (rather than dummy data!). Out-of-range data is silently ignored
Graphs are created in new windows so that they can be compared
The Supervisor's page now offers options to draw typical comparative graphs
1346 lines of code

To think about

Support graphs with "member force" y axes.
Do the readings when weights added and readings when weights removed both need to be stored in the database?

August 25th

jml2 has given the program 2 recent tests and has provided pages of feedback. I've fixed some bugs and have improved/added features. Progress so far

Data output routines made more flexible so that P, Q, or R can be used
More graphics options available
Tabbing control better
Supervisor has an option to zap a user's data
Phase 3 no longer uses phase 1's data
"all" added to some menus (in addition to phase 1, phase 1+2, and phase 3)
When supervisor's "Load cell calibration for all materials" graph is drawn for "all" phases, the datapoints for phase 3 are shown as crosses. The same happens for some user graphs.
Estimated Member Forces are no longer shown to 10 decimal places.
Code re-organised

September 12th

After a rehearsal some mods are going to be suggested by jml2. Meanwhile, we've come to a compromise on authentication. Each session's worth of students will have a password, and supervisors will have a password to an account that's sufficient to start a browser.

I've had to change several things to make the "group password" thing work. The first page (which needs more work to make it friendlier) now invites people either to login as a supervisor using their PIN, or to login as a group. I've removed mention of the term "user" where it's inappropriate.

Once people have managed to log in, things should be much as before. Note

There's now a "logout" button on the main page. It's like the hermes webmail logout button except that it times out after 4 hours rather than after a few minutes. It's not vital, but during testing it's useful.
The supervisor page shows the password (if any) that works on the current day. The idea is that the supervisor can turn up on the day of a lab, login as supervisor, look-up the current password, then start-up the student-machines accordingly. Then during the de-briefing the supervisor can bring the password to the students' attention. There's also an option to find out the password for other sessions (the passwords are just samples. I'll change them before term starts)

September 15th

Another round of revisions, and some regression testing done. Here's a brief summary of work done and work remaining.

Done
- R and Q fields added in input tables
- The graphs now have different y axes ranges depending on their purpose.
- Graphs are bigger and have bigger texts
- Calibration factor shown on some graphs
- An extra "forces v Q" graph added
- Users can view the same graphs as the supervisors can
- Re-organised supervisor page:more presentation-centered - admin features are a link away
- There's an attendance list feature (when the 1A lists appear I'll add more names). The attendance page now lets you cancel the date of a student (in case you ticked them off by accident).
To do
- bigger legends
- always displaying phase 3 data in a different way from phase 1+2 data when "all" data is being displayed
- displaying calibration info on graphs
- tidy locking (currently it zaps unsaved data in other phases).
These are tedious/time-consuming rather than difficult to implement. The last item in particular risks messing up the save/load routines if I rush things.

June 2006

The 1AStatic lab uses 5 database tables owned by tpl -

Statics1AAttendance - user, datedone, number (the number of the session they're expected to attend)
Statics1AData - user, material, thegroup, phase, thecolumn, therow, value, locked
Statics1ALocking- user, material, thegroup, phase, locked
Statics1ASessioninfo - number, password, thedate
Statics1AUser - user, supervisor, thegroup, last_change

On 28th June 2006 I made copies of these. The copies have an extra "2005" on the end of their names - Statics1AAttendance2005, etc.

September 2006

Here I'll made a list of what needs to be changed each year.

Statics1AData and Statics1ALocking need to be cleared except for the data for group1 in Statics1AData - that's sample data
Statics1ASessioninfo needs to be filled with the dates of each session, and a password for each session
Statics1AAttendance - can't be filled until the labgroups have been determined. "~tpl/bin/yearlists 1astatics" creates the SQL commands needed.

And here are some modifications I made on 5th September.

http://www.eng.cam.ac.uk/to/1AStatics is a short-cut to the 1AStatics page
I've added this year's session dates and passwords. Passwords are more memorable
On the main page I've added "For help with adding data/graphics from this experiment into a report, see the hardcopy tips page.". I've put some info on that page, but not enough.
During mich2005 I changed the input pages so that the photo at the top of the page was different for each material. I've now changed the table-header color according to the material as well.

I also removed some unused code (code in the student-code that only supervisors use, and vice versa). Other mods were suggested. Here are my short-hand replies

I've not looked at the attendance stuff yet
Adding another line of input to stage 2 means tweaking various bits of code, and means that last year's data won't load in - i.e. things might go wrong unless I'm careful
The option to display the averages for each frame might already be there.
The extrapolation mod might already have been done.
I've not looked at the details of making the data non-modifiable. The underlying mechanism is there, and it may well be that supervisors already have the power to un/lock a day's values.

Several fixes were made at the end of September

suppression of out-of-bounds line-drawing was wrong - if the first segment of a line was out-of-bounds, future segments weren't drawn either
load cell calibrations for phase 1 were wrong because I forgot that in some situations, PHP reports "1" and "1plus2" as being equal.

To-do list

Add a "remove user" option to the supervisor page
Fix the repetitions that arise in the user list (users sometimes have 2 entries - for group 0 and group 1)
in terms of the attendance would it be possible to have an additional column for comments? DONE
Fix averages DONE

October

The input data might be shown as read-only because

It's been locked by people who are doing the lab that day (an UNLOCK button is available)
the data isn't for the lab that's running on that day (no UNLOCK button is available)

I've added a "Data locking" option to the Supervisor page. For each group it shows the Locking status of the data.

Old data is by default "Locked by age". There's an "unlock" button to override this locking. If that button's used, then the data becomes "Unlocked by supervisor". There's a "lock" button which will change the state to "Locked by supervisor." Data relating to the current day's lab or to a future lab is marked as "not done".

Updated October 2006

Tim Love