Skip to main content
R
rcook1-Visitor
1-Visitor
September 22, 2015
Question

Simulate - finding average strain on a surface (strain gauge simulation)

  • September 22, 2015
  • 2 replies
  • 4492 views

I am trying to determine the average strain over a defined area, in a specific direction. The goal is to simulate the reading of a strain gauge that is bonded to my part. I can measure the min/max strain on the surface. I can also measure the strain at a point. The best I've been able to do is create a pattern of points on my surface, measure the strain at each point, then create a computed measure to average the measured strain values.

Anyone have a better/cleaner way?

    2 replies

    M
    mlindqvist1-Visitor
    1-Visitor
    September 23, 2015

    I don't think there's any better/easier way than what you are doing. Does the strain vary considerably within the area of the gague? Perhaps 2 points is sufficient..? Or maybe measuring the value at the center of the gague location, is good enough?

    An alternative could be to measure the displacement at two locations: d1 at position x1,  and  d2 at pos x2. Then calculate the average strain as (norm(x1+d1)-norm(x2+d2)) /norm(x1-x2). (I.e. change in distance divided by original distance between the points. This would then be the strain in the direction of the line between the points x1 and x2.

    x1,x2,d1 and d2 are 3x1 vectors.

    /Mats L

    U
    unickque1-Visitor
    1-Visitor
    September 23, 2015

    If the strain varies considerably within the area of one strain gauge, then you should reconsider the location of the strain gauge (or maybe even reconsider the use of strain gauges)!

    M
    mlindqvist1-Visitor
    1-Visitor
    September 23, 2015

    Good point... put strain gagues where the stress/strain gradient is low, verify that FEA matches those measurements, then you can probably trust FEA in more critical locations/locations with high stress and/or high stress gradient, where FEA and/or measurements, might be less accurate.

    If the purpose of the strain gague measurement is to find out the forces acting on a structure, then it's always a good strategy to put gagues in non-critical locations with low gradients, preferably far away from constraints etc. that might be a source of error in the FEA model. This is more accurate.

    A funnny anecdote, having struggled with evaluating strain gague measurement, I learned that the expression "Murphy's law", meaning "if anything can go wrong, then it will" comes from someone doing strain gague measurements... 

    3
    346gnu13-Aquamarine
    13-Aquamarine
    September 29, 2015

    Hi,

    We created a small 'array' of points for a gauge and at each point created directional strain measures.

    The rest was done in a spreadsheet.

    For a small gauge length, the results from a single point were close enough which reduced the number of numbers requiring manipulation.

    Computed measures were ok but difficult to maintain. Spreadsheets easily enable one to expand the calculation.

    I would be interested to know how far apart your model and practical measurements are. There are so many sources of error: actual gauge positioning, Bauschinger's effect, adhesive, real versus model geometry, real versus model loads and constraints ... an accuracy of 1% seems quite optimistic.

    M
    mlindqvist1-Visitor
    1-Visitor
    September 29, 2015

    I used to work with evaluating strain gague measurements. Heavy machine with rotating load. One load vector component was supported by hydraulics, thus the force magnitude (but not direction) could easily be measured. When I started this work, the measurements were done "by them selves" so to speak, the comparison with FEA results could not be done because of insufficient technique/method to evaluate load direction. Rainflow count etc was done directly on measured signals. I felt this was unsatisfactory, so I developed a technique to find the load direction vs time. As result I could compare FEA results (calculated with load from measured hydraulic pressure), with strain gague measurements. For some gagues, the correlation was quite good, perhaps 2% deviation. For other gagues up to maybe 10% deviation between measurement and FEA was observed. If the calculated stress is to be used for fatigue life evalulation using an S-N curve, then a 10% error will produce a massive error in calculated fatigue life, especially in high cycle fatigue. In my case, I think the main reasons for the deviation, were some conservative assumptions regarding the location for load application and that the constraints were assumed to be moment free, which is not entirely correct. I didn't stay long enough at that company to refine the FE-model to improve these results. Instead I settled for developing a technique that was "good enough" for comparisons. I basically adjusted the SN curve to compensate for errors in the stress calculation, in order to obtain the "correct" expected fatigue life for a "normal" or "average" load spectrum" and then I used this for comparisons, "what if studies" etc. So eventhough the method is uncertain, I was able to answer some questions that until then had been unanswered. Long story short: +/-1% is very optimistic I would say.

    M
    mlindqvist1-Visitor
    1-Visitor
    September 29, 2015

    I found some pics from the project mentioned above. It's a series of gagues in various locations around the machine. It's measured stress vs stress calculated in FEA using measured hydraulic pressure.  Some gagues show a quite good, others not so good correlation betwen measurement and FEA. Its a machine with a load that rotates at fixed speed, but the load magnitude varies cyclically but eratically with the same frequency as the rotation. One of them is erroneous I Think, "BS5".  In most cases, the calculated stress is higher than the measured one, due to some conservative assumptions in the load application in the FE-model.

    Capture.PNG

    Terms & ConditionsAccessibility statement