Engineering & Technology: Error cataloguing

What you'll be able to do: Keep a log that turns a term of scattered lost marks into three or four habits, and that doubles as the beginning of professional practice.

The engineering-specific categories

Errors here sort into a small number of types, and the sorting is most of the value because each type has a different remedy:

Units and conversion. The most common and the most completely fixable, Usually one habit (not carrying units through the working) and an afternoon of dimensional practice ends it.

Setup. Wrong free body, wrong control volume, wrong system boundary, wrong circuit simplification. The most expensive category: everything downstream is correct work on a wrong problem.

Assumption. Applied a formula outside its validity range. Used a steady-state relation on a transient problem. This is a boundary error, not a knowledge gap.

Execution. Algebra. Fewer than students think and the least worth studying.

Communication. Method not shown, assumptions not stated, units not carried, no comment on reasonableness. Marks lost for things you knew (article 08).

Judgement. Answer unreasonable and reported anyway. The one that matters most professionally and is barely assessed academically.

The format, with the engineering field

WHAT I DID: the specific error
WHAT I BELIEVED: the rule or assumption that made it look right WHAT'S TRUE: in my own words
CATEGORY: units / setup / assumption / execution / communication / judgement WOULD IT HAVE BEEN CAUGHT? by a units check, an estimate, a sanity check, or not at all

That last field is the engineering one and it's the most useful. It converts a log of mistakes into a list of checks you should be running: and the checks are cheap. Most engineering errors are catchable by dimensional analysis or an order-of-magnitude estimate (article 02), both of which take seconds and neither of which students do.

After a term, the log answers a question worth answering: which two checks would have caught most of my errors? Then you run those two, always.

The weekly analysis

Here's this week's error log. Group by CATEGORY, not by topic. Which category dominates? And of my errors, what fraction would have been caught by a units check or an order-of-magnitude estimate?

That fraction is usually startling (commonly more than half) and it's the argument for the habit that no amount of being told makes.

Across the disciplines

Structural and civil. Setup errors dominate: wrong support conditions, wrong load case, wrong free body. Log the diagram, not the number.

Mechanical. Units and sign conventions. Torque, RPM versus rad/s, and absolute versus gauge pressure are three recurring offenders.

Electrical. Sign conventions and reference directions. Almost everyone's log collapses to one habit about assumed current direction.

Chemical. Basis errors (per hour, per mole, per batch) which cost whole questions and are prevented by one line stated up front.

Thermodynamics. System boundary and sign convention for work and heat. Two beliefs, many wrong answers.

Fluids. Regime errors: laminar correlations on turbulent flow. An assumption error, and it produces smooth, confident, wrong results.

Software engineering. See the computing subject's article; the categories are different but the log structure is identical.

Why this one continues past the exam

Unusually for a study method, this is professional practice. Engineering organisations keep lessons-learned registers, incident logs and non-conformance records for exactly the reason this article gives: errors cluster, individual incidents look unique, and the pattern only exists in the record.

A student who keeps a personal version arrives already doing the thing the profession does, and (more usefully) already believing that recording your own errors is normal rather than shameful. That belief is not universal and it's worth having early.

Pitfalls

1. Skipping "what I believed". Makes it a list of corrections.
2. Not categorising. The category determines the remedy.
3. Skipping the "would it have been caught" field. That's the one that produces the habit change.
4. Logging execution slips. Only if repeated, then they're not slips.
5. Never acting on the dominant category. If units dominate, spend the afternoon.
6. The tell: a term of entries and you're still not carrying units through your working.

Try this today

Take your last five lost marks. Categorise each: units, setup, assumption, execution, communication, judgement.

Then ask the important question: how many would a units check or a thirty-second estimate have caught?