Module 6 - Measurement Uncertainty

Question 1

What are some reasons for the calibration of an instrument?

Answer 1

• determine the input-output relationship (calibration curve)
• Compare the instrument to a more accurate standard to ensure accuracy
• ensure zero-drift and sensitivity drift are accounted for over time

Question 2

What is accuracy vs precision?

Answer 2

Accuracy: how close a set of measurements are to the true value
Precision: agreement of the measurements (how close they are to each other)

Question 3

What is repeatability vs reproducibility?

Answer 3

Repeatability: agreement of values measured under the same circumstances (same person, way, equipment, place)

Reproducibility: agreement of values measured under different circumstances (different person, way, equipment, place)

Question 4

What is Type A uncertainty? Type B uncertainty?

Answer 4

Type A: uncertainty derived from repeated measurements around- evaluated using statistics

Type B: uncertainty from other sources (specs sheets, calibration, previous experiments, common sense, published info) - evaluated assuming rectangular distribution

Question 5

What distribution is random uncertainty assumed to follow?

Answer 5

Normal (Gaussian)

Question 6

How is the combined standard uncertainty from individual uncertainties calculated?

Answer 6

Root sum of squares

Question 7

How would you determine the uncertainty in density from the ideal gas equation?

Answer 7

1. Determine standard error in each measured parameter (T, P)
2. Determine sensitivity coefficients for each parameter
3. Apply root sum of squares to sensitivity coefficients multiplied by the uncertainty in each parameter

Question 8

Why is an over-prediction of uncertainty undesirable? Why is an under prediction also undesirable

Answer 8

Over-prediction: may lead to purchasing an unnecessarily expensive piece of equipment, or may lead to a missed physical effect

Under-prediction: false sense of security in results, overconfidence in results

Question 9

Why should uncertainty be estimated when designing an experiment?

Answer 9

• identify right equipment for the experiment
• determine if the experiment will observe a significant effect (can you observe the effect you’re studying)
• improve the experiment in the design phase
• may indicate a different approach (ie different variables) are needed to measure the effect

Question 10

What are tolerances? Why are they not uncertainty?

Answer 10

Tolerances are accepted limits chosen for a part or process. They are not uncertainty as they are accepted limits, while uncertainty is the doubt we have in our measurements