Download as pdf or txt
Download as pdf or txt
You are on page 1of 30

EML 3301

Mechanics of Materials Laboratory

Lab 2 Procedure, Discussion, Calibration


Uncertainty
The following expands on the lab 2 instructions.
If conflicts occur, the lab instructions are to be
followed.
Lab 2: Procedure, Continued

5. Weigh the unopened can with your scale 10 times. Each


time, remove and place the can on the beam. We are
examining the repeatability of placement. Find or develop
an estimate of the weight of the beverage using
references or an available scale and compare.

– Later for discussion you will:


• Take a mean
• Develop a 95% confidence interval
• This uncertainty will be used to discuss the effect your placement
has on the uncertainty in measurement with respect to the
uncertainty in the estimate of weight (MoM or Calibration)
Lab 2: Procedure

6. Open can(if needed) and take a gulp, weigh:


let the Vi run for at least 10 data points for
each different weight (make sure any
transients have stabilized) to record weight
estimated from both processes (calibration
and strain measurement) using your bream
scale. Repeat gulp and weight until empty
(let VI run throughout this step). Stop VI and
export to excel when empty.
Lab 2: Procedure
7. Weigh the empty can on your scale.
Lab 2: Discussion
1. Estimate the maximum weight that can be measured by the scale (center of mass at location detailed in Fig 2) if the
material is 1095 steel without permanently changing the beam. Assume that the gage is not a limiting factor. This should
ideally (but not necessary) be done before the use of the beam as a scale to keep from weighing too heavy of an object.
2. Report the applied weight of the full can found with two ways: (1) From the measured tared voltage deltaVamp and your
calibration curve, and (2) from the computed stress. (This is a simply supported beam for which you should be able to
compute the surface stress as a function of load and geometry, and conversely, the load from the stress.)
3. Estimate the uncertainty of both weight measurements. For the first method, consider the uncertainty in the calibration
weights, Y, and the uncertainty in the delta Vamp, calibration voltages, X. Assume that these uncertainties are reflected in
the slope that you find in the linear fit. Propagate accordingly. For the second method, consider the uncertainty in the
voltage measurements, material modulus, gage factor, and relevant measured lengths. Compare and discuss these two
uncertainties.
4. Calculate and report the mean and standard deviation of the weight found from the placement of the full can 10 times (in
lab step 8) based on one of the weight measurement methods. Select AND NOTE the method based on a quantitative
argument developed in discussion issue 3.
5. Use the Student’s t-distribution, 95% confidence interval, to estimate the statistical uncertainty of the weight of the
unopened can (In lab, step 5). Report and discuss in light of the uncertainties found in 3. Is repeatability of can
placement an issue? Use the results of this step to discuss the effect the placement might have on the overall
uncertainty.
6. Determine and report the mean and standard deviation of your gulp size based on one of the weight measurement
methods. Select the method based on a quantitative argument concerning uncertainty developed in discussion issue 3.
7. Develop/find a reference for the weight of the empty vessel used (can). Discuss any variation found in determining the
weight of the empty can with the beam scale and reference weight.
8. Recommend improvements to reduce the uncertainty for the instrumented simple beam.
9. Does the weight of the beam affect the calibration? Explain.
Lab 2: Discussion
1. Estimate the maximum weight that can be measured by the scale
(center of mass at location detailed in Fig 2) if the material is 6061
T6 aluminum without permanently changing the beam. Assume
that the gage is not a limiting factor. This should ideally (but not
necessary) be done before the use of the beam as a scale to keep
from weighing too heavy of an object.
– Use the geometry you used to weight the beverage.
– Use the material properties of the beam
– Do not consider the strain gauge strain limit, only consider yielding
of beam.
– Do not just quote the value in the report, explain it’s origin, and if
the yield of the beam will be the limiting factor for use of the scale.
Lab 2: Discussion
2. Report the applied weight of the full can
found with two ways: (1) From the measured
tared voltage deltaVamp and your calibration
curve, and (2) from the computed stress. (This is
a simply supported beam for which you should
be able to compute the surface stress as a
function of load and geometry, and conversely,
the load from the stress.)
Lab 2: Discussion
3. Estimate the uncertainty of both weight
measurements. For the first method, consider the
uncertainty in the calibration weights, Y, and the
uncertainty in the delta Vamp, calibration voltages, X.
Assume that these uncertainties are reflected in the slope
that you find in the linear fit. Propagate accordingly. For
the second method, consider the uncertainty in the
voltage measurements, material modulus, gage factor,
and relevant measured lengths. Compare and discuss
these two uncertainties.
• This step should allow the selection of one of the
methods to finish the discussion items with.
Lab 2: Discussion
4. Calculate and report the mean and standard
deviation of the weight found from the
placement of the full can 10 times (in lab step 8)
based on one of the weight measurement
methods. Select AND NOTE the method based
on a quantitative argument developed in
discussion issue 3.
Lab 2: Discussion
5. Use the Student’s t-distribution, 95%
confidence interval, to estimate the statistical
uncertainty of the weight of the unopened can
(In lab, step 5). Report and discuss considering
the uncertainties found in 3. Is repeatability of
can placement an issue? Use the results of this
step to discuss the effect the placement might
have on the overall uncertainty.
• This is an oom comparison.
Lab 2: Discussion
6. Determine and report the mean and standard
deviation of your gulp size based on one of the
weight measurement methods. Select the
method based on a quantitative argument
concerning uncertainty developed in discussion
issue 3.
Lab 2: Discussion
7. Develop/find a reference for the weight of the
empty vessel used (can). Discuss any variation
found in determining the weight of the empty
can with the beam scale and reference weight.
Lab 2: Discussion
8. Recommend improvements to reduce the
uncertainty for the instrumented simple beam.
– Try to be quantitative.

9. Does the weight of the beam affect the


calibration? Explain.
– Tare…
Lab 2 Uncertainty in Calibration Slope
• Linear fits do not lend themselves to the RSS
approach
• Instead, we will use a set of simulations, and
look at the variation in the slope as a function
of the variation of the input (the deltaVamp
and the calibration weights)
Lab 2: Calibration Uncertainty
• VI: strain gage as a transducer with calibration
– Estimate the weights of calibration volumes of
water
– Measure one analog voltage
• (DeltaVamp is the difference between two measured
Vamps)
– Only consider the slope, and only consider a
change in weight on the beam.
– After calibration, apply calibration expression to
generate weight from voltage reading in VI.
Lab 2: Calibration Uncertainty
• Calibration
– Assume deltaVamp is linearly proportional to the load applied to
the beam (previously shown…). Then the following relationship
holds:
𝑦 = 𝑚𝑥 + 𝑏
with y = weight, m → slope, x → Vamp and b → intercept
• The calibration procedure is finding m as best we can to allow a
weight to be measured.
– b can be dealt with by measuring a change in voltage for a change in weight
• Least Squares is a suitable approach to find m.
– Excel will preform a least squares using the “trendline” function on a graph.
• Each data point has some uncertainty.
– Error bars are useful for visualizing the magnitude of this error
Lab 2: Calibration Uncertainty
• How can we estimate the uncertainty in the slop of the fit, m?
– Find the uncertainty in each data point
– Simulate a new set of data that is appropriate with respect to the
uncertainty
– The simulated set should:
• Have a Gaussian distribution
• Be random
• A group of simulated data sets should have a similar mean as the measured
experimental data
– Generate a slope for the simulated data set
– Repeat until sufficient simulated m’s are known to allow a standard
deviation to be developed.
– Assume enough simulated slopes are generated to allow a true
standard deviation to be estimated.
– Use twice the standard deviation as the uncertainty in m (assuming N
→ infinity for our purposes).
Lab 2: Calibration Uncertainty
• Find the uncertainty in each data point
– Estimate the uncertainty in calibration weights
– Estimate the uncertainty of Vamp from the SADI
DAQ.
Lab 2: Calibration Uncertainty
• Find the uncertainty in each data point
– Weights
• If each volume is measured separately, quantify the
accuracy of the measurement
– Use 2% is nothing else.
• If volumes are measured and added, use RSS and
uncertainty in each measurement.
Lab 2 Uncertainty in Calibration
Weights
Lab 2: Calibration Uncertainty
• Find the uncertainty in each data point
– Propagate uncertainty
𝑊𝑡 = 𝑉𝑜𝑙 ∗ 𝜌 ∗ 𝑔
• Where
– Vol → volume of water, ml
– Ρ → density of water, g/ml
– G → acceleration due to gravity, m/s2
Lab 2 Uncertainty in Calibration
Weights (excel example)
• RSS: 𝑊𝑡 = 𝑉𝑜𝑙 ∗ 𝜌 ∗ 𝑔
– 𝜇𝑉𝑜𝑙 ≈ 2%
– 𝜇𝜌 ≈ 0 ?
𝑚
– 𝜇𝑔 = 0.01
𝑠2

𝜇𝑊𝑡 2 = 𝜇𝑉𝑜𝑙 𝜌 𝑔 2
+ 02 + 𝜇𝑔 𝑉𝑜𝑙 𝜌 2
Lab 2: Calibration Uncertainty
• Uncertainty in deltaVamp
– Uncertainty in each measurement
• Store standard deviation in VI
– Select largest one over range (unless truly outlier)
• Use 95% CI for Vamp
– 𝑑𝑒𝑙𝑡𝑎𝑉𝑎𝑚𝑝 = 𝑉𝑎𝑚𝑝 − 𝑉𝑎𝑚𝑝𝑡𝑎𝑟𝑒
• Where
– Vamp → measured Vamp for calibration weight
– Vamptare → tare value selected to tare out bias errors
• Propagate uncertainties using RSS
Lab 2: Calibration Uncertainty
• Uncertainty in deltaVamp
𝑑𝑒𝑙𝑡𝑎𝑉𝑎𝑚𝑝 = 𝑉𝑎𝑚𝑝 − 𝑉𝑎𝑚𝑝𝑡𝑎𝑟𝑒
• RSS

𝜇𝑑𝑒𝑙𝑡𝑎𝑉𝑎𝑚𝑝2 = 𝜇𝑉𝑎𝑚𝑝 2
+ 𝜇𝑉𝑎𝑚𝑝𝑡𝑎𝑟𝑒 2

– Assume 𝜇𝑉𝑎𝑚𝑝 = 𝜇𝑉𝑎𝑚𝑝𝑡𝑎𝑟𝑒


Lab 2: Calibration Uncertainty
• Simulate a new set of data that is appropriate
with respect to the uncertainty
• The simulated set should:
– Have a Gaussian distribution
– Be random
• Generate a slope for the simulated data set
Lab 2: Calibration Uncertainty
• Repeat until sufficient simulated m’s are known
to allow a standard deviation to be developed
(1000’s+).
• Assume enough simulated slopes are generated
to allow a true standard deviation to be
estimated.
• An excel sheet is provided, linked under the Lab 2
assignment, that does this:
• Use twice the standard deviation as the
uncertainty in m.
Lab 2: Calibration Uncertainty
• Given an uncertainty in the slope, we can estimate an
uncertainty in the weight by using:
2 2
𝜇𝐶𝑎𝑙_𝑤𝑡 = 𝜇𝑠𝑙𝑜𝑝𝑒 𝑉𝑉𝑑𝑒𝑙𝑡𝑎_𝑎𝑚𝑝 + 𝜇𝑉𝑑𝑒𝑙𝑡𝑎_𝑎𝑚𝑝 𝑚

– Where
• μslope = uncertainty in slope from fit
– Monte Carlo Simulation
• μ Vdelta_amp = uncertainty in tared voltage from AD in SADI DAQ
– RSS
• VVdelta_amp = measured tared voltage
• m = slope from fit
• It is left to the student to address the uncertainty in b, the
intercept ( not needed if only changes are considered)
Lab 2: MOM Uncertainty
• VI: strain gage to measure strain, estimate
stress and load
– Measure some geometry
– Find some material properties
– Measure two analog voltages
– Apply a measured zero value to Vamp and scale
(gain) to get Vg
– Weight things
Lab 2 Uncertainty Strain based
• Use RSS to propagate the uncertainty in the
measurements into the uncertainty of the
load
More Background, Gain Windows

You might also like