Activity Feed › Discussion Forums › Strictly Surveying › Movement Monitoring equipment/method query
Movement Monitoring equipment/method query
Posted by benchmarkses on June 18, 2020 at 8:38 pmHi All
I run a small site engineering company based in London, primarily doing setting out and some surveys. In recent months I’ve been getting more and more monitoring jobs (all buildings), so now I have around 10 active monitoring projects on the go.
First of all, I’m using a TS15, because that’s been fine for setting out and surveys. Before you start berating me on this, it’s a 1″ R1000 so pretty good spec. However, I have started running into some issues with the monitoring results and would like to get some feedback from you more experienced monitoring guys.
As an example, I have one job where they want the front and the rear of the house monitoring. The front I put my reference targets as far away as possible, usually four or five of them just in case, and in general the results are ok. I do however sometimes get 2 or 3mm movement ‘sideways’ which I know is not possible – is this caused by angular error in the instrument and if so is there any solution? I do use two face measurements. The rear is a different story. Small garden, my reference points are a similar distance away from the instrument as the monitoring targets simply because I either can’t see any further outside the garden or don’t have access to install targets on neighbouring properties. The results have gone a bit haywire, showing movement that is not likely at all like 6mm sideways and 10mm up. So I’m now left in a position where I have another 6 months of monitoring to do and getting unreliable results already after a few visits.
So my questions are:
1. Is it the instrument, do I need a TS60 or similar?
2. How do I achieve greater reliability in general on monitoring projects?
3. How do reliably monitor in more confined spaces such as a back garden of 10m x 20m?
(apologies if this is poorly written, it’s late and I’m tired :))
benchmarkses replied 3 years, 8 months ago 13 Members · 30 Replies- 30 Replies
What are your targets, prisms?
Could be centering errors check and adjust all tribrachs, optical plummets etc.
It’s retro targets so I’m sighting them, and doing two faces as I said before.
Good idea to check the optical plummet, I will do so.
Are you certain that your control points are stable?
What type of material (ie. vinyl, wood, steel) have you attached your targets to and have you considered the affects of expansion and contraction due to temperature variations?
I like to begin a monitoring project with a loop closure on a large network of intervisible control, then adjust it in Star*Net. I’ll then use the adjusted coordinates for resections between around five control points. This eliminates the problem of centering errors.
I monitor a lot of dams (39). Some phenomena I see:
1) on lock walls (long linear walls), we separate the movement into parallel to the long axis and perpendicular. The only one that really matters (most of the time) is the perpendicular. We can see 3 cm or more of difference along the long axis depending on temperature. The way these are constructed is there is a lock chamber (360 to 1200 feet long) with lock gates at the ends of the chamber. The gate monoliths are usually constructed down to bedrock whereas all of the other monoliths are typically on piles. So the gate monoliths constrain the along wall expansion in the chamber. Then there are upstream and downstream extensions from the gate monoliths, sometimes 1000 feet or more long. These definitely do expand at the unconstrained end.
2) We run levels to the interior of the intake towers (concrete) to the water level gages. There is a noticeable difference between winter and summer, also due to temperature (my theory)
I bring these up to show that temperature can be a factor.
Great responses so far. The critical question, as John touched on, is what kind of movement are you looking for, and how much?
There needs to be clear direction on how much movement is expected or tolerated. The general rule of thumb is to set up your observations so that the precision of your final coordinates for monitoring points is one-third that of the smallest expected movement.
In order to get that, the control network has to be a lot tighter than that. As a very general rule, you want your control network to be at least twice as precise as your desired accuracy, and ideally closing in on an order of magnitude better. So if you want to observe to the nearest centimeter, your control network needs to have a relative accuracy of five millimeters or better, and ideally one millimeter. That can take a lot of work to establish.
Unless the instrument is out of adjustment, a one-second total station is just fine for most monitoring projects. Proper procedures and number of redundant observations are what really make the difference between a quality monitoring session.
You do need to ensure that for each point you are observing twice in each face in order to obtain that 1″ precision. As Murphy said, you are going to want to resection / free station for each monitoring session, with a zero instrument height to eliminate centering and measure-up errors.
I highly recommend Star*NET or another program that can run a preanalysis to determine what kind of precision you can expect from a typical session. Otherwise you really won’t know whether the coordinates you are seeing are indicative of actual movement, or propagated error. It will really help you see the effects of short sights and show you how to compensate, whether that is more setups, more control, or a different instrument.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil PostmanI suggest that a house, or any other building, will move a measurable amount from day to day, and within a day/night cycle. The variation you are seeing is in range. 6mm is getting to the outside of that range, probably, but still not outrageous.
Monitoring can be a real education about the accuracy/precision of measurements.
Thanks for the replies all.
I’m monitoring generally residential houses so nothing as big as a dam. I wouldn’t think that a brick house would be subject to a great deal of movement due to temperature, but hey I could be entirely wrong. Targets are attached to brick or concrete using RMSK adhesive.
I haven’t done traverses since college (never felt the need at work), but maybe I need to be adjusting my control stations. Even though my resections are generally pretty good (0 to 1mm errors). Unfortunately I can’t check the angular error because I’m now using Xpad.
A colleague has just suggested that I use fixed prisms for my control points and set up using sets of angles – does this make sense to you?
Sets of angles, definitely. Fixed prisms seem unnecessarily expensive, but would improve precision, for sure.
Have been taught to use prisms and never reflector targets. Sighting errors due to different users at different epochs may cause additional errors? Just a thought…
As others mention. I use set collection (set of angles equivalent on carlson) and run at least 4 sets and check the residuals of the angles. Generally the angles are tight and when back in office drop it into starnet to check the results and plot into graph.
Before starting any monitoring, we do a check and adjust to confirm the values of the TS.
There was a fire under a major bridge here in 2017. It damaged one of the main beams. Almost caused the structure to fail. So the DOT had one of their local survey contractors monitor the bridge. Here is what it said in the paper…
“A crew used a surveying device known as a robotic total station. The device uses mirrors, a laser, and a series of sensors to test every half hour 24 hours a day for any movement of the bridge…The good news is the bridge has moved less than 1/1000th of an inch since the fire”.
That statement made my head hurt…Total 100% lack of understanding by the company that was taking the measurements. Or maybe they were misquoted…but somewhere they got that 0.001″ figure. even if it was 0.001′, not possible.nor 0.001 m. Maybe 0.003 m, but you would have to take into account the expansion/contraction due to temperature.
- Posted by: @sireath
Have been taught to use prisms and never reflector targets. Sighting errors due to different users at different epochs may cause additional errors? Just a thought…
Ideally yes, but prisms are usually more expensive and more difficult to mount. We will keep the same crew with the same gear on a monitoring project for its lifespan; however, there are inevitably days when that crew is tied up and they have to trade off their gear to a different crew, which will then do the measurements.
With reflective targets I will definitely see a slight difference between the two crews (a very minor “bump” in the trendline), but with experienced crew members and a tight control network the differences are rarely statistically significant. Some total stations can recognize reflective foil targets with ATR as well.
Posted by: @john-hamiltonThat statement made my head hurt…Total 100% lack of understanding by the company that was taking the measurements.
I am consistently amazed at the clients who tell us that their previous surveyors were “able to measure to one-thousandth of an inch“, which leads to “Why can’t you do the same thing? Maybe you should get [basic surveying instrument XXXXX] because that’s what the other firm used. They told us it’s the most accurate instrument around.”
If you have to explain basic statistics and error propagation to a client, you’re already starting out behind the curve…throw in temperature variations and it’s a whole another level of difficulty. Once they have been conditioned to think that surveyors can get subatomic precision it is tough to convince them otherwise.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil Postman Monitoring is simply screwing up measurements exactly the same way every time. Take the same measurements with the same equipment under the same conditions when possible. Report your results in a way your client can use them. They normally have no use for coordinate relationships, but prefer upstream, downstream, etc..
I am taking the same measurements from the same position using the same control points and using the same instrument, that’s why it’s so frustrating when I start getting 2-3mm errors……
If you analyse those ties with LS, with reasonable a priori error estimates for the equipment used, you will find that 2-3mm is well within acceptable limits.
@benchmarkses They could be errors and also they could be movement due to expansion and contraction of the building. You have to isolate to be really sure that they are errors. Even in tunnels with automated monitoring stations, we were getting 3-5mm and 20-30mm at regular intervals. Took us awhile to connnect the dots that it was due to trains passing by in the tunnel.
Are retro/reflective targets really good enough for monitoring? I have never used retro reflective targets but I imagine if you are not measuring perpendicular to the target it is just the same as a reflector-less measurement, in that the angle of incidence can affect accuracy.
That is true, but as @thebionicman said above, the idea is to screw up the measurements the same way each time so you have a valid comparison.
If you are setting up in the same place each time, with the same angle of incidence, the returns will be the same.
“…people will come to love their oppression, to adore the technologies that undo their capacities to think.” -Neil PostmanIn my opinion, the most common problem the prism constant.
We do a lot of monitoring (BenchmarkSES will know of the Metrocentre). It is unusual but quite possible to get a building which floats around sideways, sometimes on a diurnal basis (always do your obs at the same time each day), sometimes on water content of the soil, sometimes for other factors such as tidal – that can spread for several hundred metres beyond the river (in which case do the obs at the same state of tide AND record tide height. (Oh, yes and the 11 year sun-spot cycle can throw in to leave the odd residual 1mm that cannot be explained).
One thing I have learnt is that it is very wise to fix several reference points well away from the site, so that you can observe these as a separate check to see if your stations are moving relative to those. Another thing is NEVER to rely on a resection – they are the best way to introduce untraceable errors.
As long as you are observing reasonably square to the target retro/reflective should be good enough – yes, there will be a residual error, but from the same observation angle it should always be the same, so comparisons between reading sets will give a true movement.
I’ve got to disagree with your position on resections. If done well with good geometry they can be a way of eliminating errors. I resect to at least 3 points of offsite control – of the same type you seem to be using as check ins. Check into a forth.
Log in to reply.