Engineering & Technology · Sketches, Models & Technical Communication
Design notebooks, reports and communicating uncertainty
A colleague receives a file called “tank_final_new2” and a message saying “the model works”. There is no equation, input file or description of the comparison. Even a correct number is hard to use in this form: the next person cannot tell which tank, which calculation or which meaning of “works” belongs to it.
Use units and simple arithmetic from Measurement, Units & Engineering Evidence. This chapter introduces its own coordinates, projections and model vocabulary; multiplication and division are enough for the worked cases. The geometry, system, data and results are original teaching constructions. They are not observed experiments, build specifications or operating instructions. No conformance to a drafting, piping or NASA standard is claimed.
Back to the chapterAfter reading, you can
Create a technical record that lets another person identify the version, reproduce the reasoning and see what remains uncertain.
Let a notebook preserve the path of the work
A design notebook can be paper or digital. Its central job is to retain dated observations, sketches, calculations, choices and unresolved questions in a form that can be found again. For our teaching model, begin an entry with the question: estimate stored volume over 0–20 s under specified inflow conditions. Identify the author, date, model revision and comparison-record identifier.
Keep inputs and raw observations separate from calculated results. “Final listed comparison volume: 4.84 L” is different from “calculated difference: −0.16 L” and from “investigate possible flow variation”. The first describes an entry, the second a calculation, the third a proposed action. In this lesson all those values remain constructed examples, so the record must say that too.
Keep each result attached to the rule that produced it
Model R1 uses V = 1 + 0.20t. Run A evaluates that rule at the five stated times. Comparison C1 supplies the separate illustrative volumes. Difference record D1 combines A and C1. The figure shows these dependencies: changing a model input means generating a new result, not silently attaching old output to new rules.
The candidate R2 changes the rate to 0.192 L/s after fitting the C1 endpoints. Mark the reason and the data used. R2’s new calculations are distinct from R1’s results, and a later comparison should identify whether it was kept separate from fitting. A revision identifier is useful only when it resolves to the actual retained content.
Report uncertainty where it affects the conclusion
A short report should explain the question, method, key results, limitations and proposed next action. Include the equation, units, initial value, time range and comparison method where the reader can inspect them. A figure helps show the growing difference, while its text equivalent or table preserves exact values. Keep a readable export alongside the editable source and the data needed to reproduce it.
Avoid “validated” without saying for what use, against which evidence and with what remaining limitations. Report numerical uncertainty when a justified estimate exists; otherwise identify what is unquantified. A reviewer’s name alone is not a review: retain what they inspected, what they found and what happened to the findings. Corrections should leave the earlier entry recoverable and state why it changed, so the record explains the reasoning instead of merely presenting the latest answer.
Replace an unsupported claim with a useful report paragraph
Original: “The tank simulation is accurate. Final value 5.” This hides units, time, revision, comparison and meaning of accuracy.
Revised: “In the constructed teaching case, model R1 predicts 5.00 L at 20 s from 1.00 L initially and a constant 0.20 L/s inflow, with no outflow or overflow. Illustrative comparison C1 lists 4.84 L, giving comparison minus prediction = −0.16 L. The assumed ±0.05 L final-reading bound does not include the prediction. Other input and model uncertainties have not been quantified.”
Then state the decision: “Use this result to plan further checks of flow, timing and possible losses. R2 fits C1’s endpoints and awaits a separate comparison.” Attach the retained equation, five-row data table, plot and calculation record. This is enough to challenge the reasoning without mistaking the exercise for an observed tank experiment.
Try the reasoning
Why keep R1 after preparing R2?
Show the answer and reasoning
R1 explains the original prediction and disagreement. Its retained content makes the reason for R2 inspectable and prevents old outputs being mistaken for results of the new rule.
Does a dated screenshot make the calculation reproducible?
Show the answer and reasoning
It may preserve appearance, but usually omits inputs, equations or editable data. Retain those items and their version links as well.
What should replace “all uncertainties are small” when only the hypothetical reading bound is given?
Show the answer and reasoning
State the given ±0.05 L reading bound and say which other input and model uncertainties remain unquantified. Do not generalise one bound to every source of disagreement.
What to carry forward
A useful technical record connects the question, version, inputs, method, evidence, uncertainty and next action so another person can follow the work.
References
Teaching resources and research supporting this lesson’s explanations and examples.