Archie_M_notes

A new Demo Pack

Version 2.3.1 has been a long time in the development but we think it is ready for use now. We will circulate it to users as soon as we have finished testing. In the mean time, the Demo can be downloaded from.
Download Archie_demo_pack.zip (1607.6K)

Bill

Extracting Live load Thrust

I was asked at a seminar last week how to extract the live load thrust from Archie Output.
A spreadsheet to do it is here.
This is a 4 page spreadsheet. The first contains the tabular output from a dead load only case. The second has the live load case. The third does the computation and the fourth the plotting. If you want to just use this sheet you will need to make sure you have the right number of segments (60 in this case).

Bill

Complex loadings

When we first began with Archie back in 1984, lots of things were much easier. All our users worked on roads and road vehicles were pretty consistent. The rules for distribution meant that we could deal with full axle loads and distribute them over a width of 3.3m plus the depth of fill but limited to the width physically available. We called that the lane width and put in a default value of 2.5m as this is near enough the physical width of the vehicle.

Then came railway work and lots of things changed. The transverse distribution rules are different, not least because the railways don't allow the smoothing out of centrifugal effects etc. Also, the railway patch loads are rather different, under the sleeper they are 800mm wide and 250 front to back instead of 300 square.

We have made a number of modest changes to accommodate  these differences, but they require  a degree of  involvement from the user. For example, on railways we suggest using wheel loads and a lane width of 1.5m minimum. The "axle" for the wheel load needs to be 500mm long and the 300mm patch each end increases that to 800.

HB vehicles need a bit of thought too because they are 3.5m wide, so wont fit on the default 2.5m lane (what happens if you don't change it is that the load goes on the 2.5m width anyway).

So we made some more changes. In version 2.3 of Archie-M we changed the title "Lane Width" to B
"bridge Width". It is still a bit confusing, so if you can think of a better short title we would be pleased.

In the load file we now include a track width (between wheel centres on an axle) and a minimum lane width. The program now checks and wont attempt to put a load on a bridge that isn't wide enough.

Finally we get the SV vehicles which are very wide. The wheels are 2.65m apart on the axles so at 1.5+h effective width, h has to be 1.15m before these patches overlap. The model loads (which were created by a user) deal with wheels rather than axles and the track width is set to 0.1m to mimic the wheel patch. In version 2.3 it will be necessary to think about an appropriate bridge width and minimum lane width. Ideally, the minimum lane for the wheel would be 3m wide. ie symmetrical about the wheel and generating no overlap. If the bridge is narrow, or the load might get close to one edge, it may be necessary to reduce these values to account for the unavasilability of the symmetrical load path.

Life gets really complicated with the SV TT vehicle which includes the tractor and a trailer with 4 wheels per axle. There is no simple model for this case. It may even be necessary to deal with it by combining the two wheel loads in each half of an axle. Thus, with a 3.7m wide vehicle, one might have a lane width of 1.85m and track width of 0.1m where there are two wheels per axle and 0.96 where there are 4.

By the time we get to version 3, which might still be a year away, we hope to be able to deal with explicit load patterns and a sensible distribution model.

In the mean time, if you have a complex case, I have tools that might help

Bill

Demo download

We have been having difficulty with the demo page on our main web site. The latest Demo can be downloaded from here. We are aware of some bugs relating to the lane width and bridge width but if you don't try to change those the program should run OK.

The download is only 800kb and includes various bits of documentation. If you unzip the contents and leave everything in the same relationship, it should run, even from a CD.

Bill

Railway Loading

The system for applying railway loads within Archie-M is a little complex because it was added in a hurry. There are a number of separate isses that must be dealt with:

1) On a railway, the two wheels on an axle are considered to carry potentially different loads as a result of centrifugal effects etc. On roads, these effects are effectively eliminated by the distribution rules. It is therefore necessary to consider single wheels rather than axles.

2) The single wheel is distributed to a patch 800mm wide under each of three sleepers. The distribtuion to sleepers is done in the load file. The 800mm width is created by representing the wheel as a very short axle 500mm long. with a 300mm patch under each "wheel" that creates a patch 800x300 which is close enough to a sleeper patch.

3) The catual distribution is different. There is a side slope of 1:1 instead of 1:2, and there is no addition (the 1.5m in highway loading) This is dealt with by choosing railway distribtuion from Tools/Set Distribution Mode.

4) With an wheel load considered instead of an axle, the maximum distribution represented by the lane width must be reduced. The conservative approach is to assume a lane width equal to the distance between load points on the rails (usually assumed to be 1500mm. You may increase this if the rail concerned is more than 750mm from the edge of the bridge or from the nearest rail of the adjacent track.

Multi span assessments

This has come slower than any of us might have hoped:

What stabilises a slender pier in a viaduct. Obviously, as displayed within Archie, the adjacent span provides propping, but this is a very conservative view. It is impossbile to push a slice of a pier over. any live load effects will be distributed over the full width of the pier, or nearly so. If the distribution doesn't reach the full width at the pier head it will continue down the pier so the disturbing effect per metre width will progressively reduce.

Perhaps more importantly, the spandrels and parapets, even if detached from the arch, provide an effective prop from pier to pier. If the spandrels are completely free from the arch, the limiting force will be controlled by friction at the pier head, but it will still be large.

When assessing the multi span aspect of a viaduct, the "lane width" should be altered to the tota width of the bridge divided by the number of vehicles that can fit side by side. The result will still be conservative becasue there are alternative load paths as described above.

Distribution and multi spans

Getting a new handle on distribution back in November was an important step in understanding arches generally, but multi spans in particular. More is needed than I can place here now, but it is clealy impossible for live load in one span to push a strip of a pier over. The whole of the next span, plus its spandrel walls, will be mobilised to provide support. The distribution model must take account of this.

In the short term, it is necessary to fudge the issue in using Archie-M. I will add some hints on how to do this soon.

True shape input

Putting true shapes into Archie-M is fraught with a number of difficulties, not least of which is getting the survey right.

Some basic rules:

If the survey accuracy (not the precision!) is not better than about 20mm don't bother, fit a standard shape. If you survey using a total station, you will pass this barrier, but not if you use a tape and staff.

You need a group of survey points near each end and near any definite changes in slope. The ends are particularly important  if the shape is not approximately circular. The absolute minimum is 9 points round the arch. If they are spaced roughly equally round the curve (note not equal spaces horizontally), adding one between the end two at each end will dramatically improve the fit.

Getting the data in:

It is best to do a simple check before you input the data. put the numbers (x and y coordinates) into Excel as two columns and plot the result as an xyplot using the curve fit option. If the curve doesn't look good in excel it will not look good in Archie-M. You can often overcome this difficulty by adding extra points by eye. Adding points is much easier in Excel because you can simply insert a row. adding a row in Archie-M adds it at the bottom and you then have to move the numbers down one at a time.

Once you have two columns of numbers which plot effectively, save them as a .prn file. This option is available in the save as list for excel.

The final step is to rename the file as .sur

Once it is in that form it can be read into Archie-M.

Note that the data points should start in A1, there should be no labels and no numbering of the points, just two columns of numbers.

We are working on a new version of Archie-M and hope to improve some aspects of the data input.

Distribution in Arches

It's good to feel my brain coming back after two or more years on strike.

I have been struggling with the issues of distribution in arches for some years. It is clear that the "distribution through the fill" models currently used are just plain wrong. Results from tests done years ago show that there is precious little distribution at working load (which is what we really care about.

The arch, though, is capable of massive distribution, it's just a question of how to model it. If we assume that an arches distributes concentrated thrust in the same way as a wall would, The effect of live load in a strip declines as it flows towards the abutment (Except, of course, when thrust from another load starts to distriibute into the strip).

At last, I have found a way. The results so far are very encouraging. I think it opens a path into sensible results for skew bridges. I'll put more here when I get another chance.

New threads

I have just started a new thread on my general Arches Blog about damage to arch bridges. You will find the first post here.