START HERE → 01
1. Units: the engineering reasoning pipeline
One quantity, followed from meaning → model → engineering decision
[meaning] tag each quantity with a unit
[language] SI bases + derived units
[check] test equation dimensions
[convert] factors of 1 cancel units
[precision] report justified digits
names what the number means
reveals what quantities are
incompatible terms cannot be combined
makes equal amounts comparable
measurements limit reportable digits
Pipeline preview · from a tagged speed to a usable result
60 mi/h → meters per second
60 mi/h × (1609.344 m / 1 mi) × (1 h / 3600 s)
mi cancels mi; h cancels h → 26.8224 m/s ≈ 26.8 m/s
The conversion factors equal 1; the remaining units prove the result is a speed.
Shared language
Why units matter · MCO
SI base dimensions are our reference language. Customary units are valid too—state them, then convert explicitly. Mass (kg) is not force (N); weight is a force.
MCO was lost after thruster-impulse data used lbf·s where the receiving interface expected N·s: a unit-convention mismatch.
Takeaway: carry units through each step; they give numbers meaning and expose incompatible conventions.
meaning
language
check
convert
precision
infer
debug
Units translate a measured number into a specific kind of quantity.
Seven SI base quantities → their defining units
[infer] constrain possible formulas
[debug] diagnose unit + quantity mix-ups
Length meter (m) · L
Mass kilogram (kg) · M
Time second (s) · T
Current ampere (A) · I
Temperature kelvin (K) · Θ
Amount mole (mol) · N
Luminous intensity candela (cd) · J
models must have compatible dimensions
apply the chain to real calculations
Engineering’s frequent five: m, kg, s, A, K. L, M, T, I, Θ, N, J denote dimensions—not unit symbols.
SI prefixes · scale the unit, not the quantity type
n · nano 10⁻⁹
µ · micro 10⁻⁶
m · milli 10⁻³
unit 10⁰
k · kilo 10³
M · mega 10⁶
G · giga 10⁹
Each prefix multiplies a unit by its power of ten: 1 km = 10³ m; 1 ms = 10⁻³ s. In between: c = 10⁻² (centi), d = 10⁻¹ (deci), da = 10¹ (deca), h = 10² (hecto).
Case matters: m = milli, M = mega. Prefixes change magnitude; dimensions still say what kind of quantity it is.
Derived units: definitions build from base units
FORCE · newton (N)
F = ma → N = kg·m/s² = kg·m·s⁻²
Mass × acceleration: each newton is the push that gives 1 kg an acceleration of 1 m/s².
PRESSURE · pascal (Pa)
Pa = N/m² = kg·m⁻¹·s⁻²
Force spread over area; dividing by m² subtracts two powers of length.
ENERGY · joule (J)
J = N·m = kg·m²·s⁻²
Work transfers energy: force acting through distance gives N·m.
POWER · watt (W)
W = J/s = kg·m²·s⁻³
The same energy delivered per second is more power.
Customary units · same dimensions, different units
LENGTH: 1 ft = 0.3048 m; 1 in = 0.0254 m (exact)
MASS: 1 lbm = 0.45359237 kg (exact); 1 slug = 1 lbf·s²/ft ≈ 14.59 kg
At g₀ = 32.174 ft/s², 1 lbm weighs 1 lbf. Generally F(lbf) = m(lbm)a(ft/s²)/g_c, with g_c = 32.174 lbm·ft/(lbf·s²).
PRESSURE: 1 psi = 1 lbf/in² ≈ 6.895 kPa = 6.895×10³ Pa
“Pound” is ambiguous: lbm is mass; lbf is force. A slug is also mass. Keep dimensions explicit—mass M ≠ force M·L·T⁻².
Worked example · unit structure tells what 6 means
Given: mass m = 3 kg; acceleration a = 2 m/s²
F = ma = (3 kg)(2 m/s²) = 6 kg·m/s²
6 N
N names force: kg·m/s². The number is 6; its unit says “force.”
What units tell us—and what they don’t
Unit names and conversion factors depend on the stated system. Dimensions identify quantity type, not magnitude (2 m and 8 m are both length). Dimensional consistency is necessary, not proof a formula is correct; a wrong equation can still have matching dimensions.
Self-check · expand each named unit into kg, m, s
Try N, Pa, J, and W; then compare with the expansions below.
N
kg·m·s⁻²
Pa
kg·m⁻¹·s⁻²
J
kg·m²·s⁻²
W
kg·m²·s⁻³
Derived-unit building blocks · dimension exponents
Quantity / unit
M
L
T
Force · N
1
−2
Pressure · Pa
−1
Energy · J
Power · W
2
−3
Each row is MᵃLᵇTᶜ: the exponents match the expanded SI units above.
Unit recipes · follow the building blocks
kg·m/s²
× m
÷ s
force
÷ m²
energy
power
pressure
Multiply or divide by the defining unit: N·m → J; J/s → W; N/m² → Pa.
Keep the type separate from the scale
Dimension M·L·T⁻² = force
Unit choice N (SI) or lbf (customary)
Both labels describe force, but the number changes with the unit: 1 lbf ≈ 4.44822 N. Dimensions classify the quantity; the chosen unit sets its numerical scale.
Takeaway · build derived units from base units, then keep quantity type and unit scale explicit.
Pipeline context · units → base-language → equation check → conversion → precision → inference → debugging
[meaning]
[language]
[check]
[convert]
[precision]
[infer]
[debug]
From [language]: rewrite quantities in SI base dimensions.
DIMENSION LEDGER · equal signs must balance
Addition: [L] + [L] → [L] ✓ same dimensions
Not allowed: [L] + [T] ✕ unlike dimensions
Multiplication/division combine dimensions: [L][T] and [L]/[T]. A pure number (e.g., ½) has dimension 1, so it changes no dimension.
Like dimensions may be added or equated; unlike ones fail the check.
Worked check · distance = ½ acceleration × time²
[x] = [½] · [a] · [t]² = 1 · (L/T²) · T² = L ✓
The seconds² cancel the s⁻²; the dimension left is length.
Put in the stated SI values:
x = ½(3 m/s²)(4 s)² = ½ × 3 × 16 × m·s²/s² = 24 m
Result: 24 m has the required dimension L, so the equation passes this check.
Countercheck · the plausible-looking x = at
Claim: distance x = acceleration a × time t
Left: [x] = L
Right: [a][t] = (L/T²)T = L/T
L ≠ L/T: the right side is velocity, not distance; the equality fails dimensional homogeneity.
At these values, at = 12 m/s—not a distance such as 24 m.
A unit check rejects this equation before its numerical result can mislead.
Function gate · inputs must be dimensionless
For sin, exp, log: check the entire argument first. sin(ωt): [ω][t] = (T⁻¹)(T) = 1, so it passes; ω is in s⁻¹ and t in s.
sin(ωt) ✓
exp(1) ✓
log(1) ✓
Each shown argument is a pure number (dimension 1), not a dimensional quantity.
What a pass does—and does not—tell you
Necessary, not sufficient
x = at² and x = ½at² both have dimension L: units alone cannot choose the coefficient or prove the model/measurement correct. Nor distinguish distance from displacement (both L).
Dimension-vector check · exponents of (L, M, T)
Each column records the power of one base dimension.
x and ½at²
0
match ✓
at
not x ✕
The time exponent exposes the mismatch: velocity has T⁻¹; distance has T⁰.
Function arguments: cancel dimensions before evaluating
ωt = (s⁻¹)(s) = 1 sin(ωt) ✓
x has dimension L sin(x) ✕
sin, exp and log take pure-number arguments; if the input is dimensional, form a dimensionless ratio first.
Takeaway · matching dimensions let an equation pass; they do not prove it true or determine its constants.
What the dimensional balance cannot decide
x = at²
x = ½at²
[x] = L
[RHS] = L
Both pass the dimension test: it cannot select the ½, prove either model, or tell distance from displacement (both L).
Dimension algebra · exponents combine
Product: at²
Multiply → add exponents: −2 + 2 = 0
[a][t]² = (L T⁻²)(T²) = L
L ✓
Quotient: x/t
Divide → subtract exponents; velocity results.
[x]/[t] = L/T = L T⁻¹
L T⁻¹
This exponent bookkeeping explains why s² cancels s⁻² in the valid motion equation.
Make a dimensional input into a pure ratio
x · L
sin(x/x₀) exp(x/x₀) log(x/x₀)
x/x₀ = 1
x₀ · L
same dimension upstairs/downstairs
Each function receives a number, not a length. For angles, radians are dimensionless too.
Pipeline context · each step prepares the next
From [check]: preserve the quantity’s dimension while changing its unit.
A conversion factor is a ratio equal to 1
1 mi = 1609.344 m → same distance, two unit labels
1609.344 m ──────────── = 1
1 mi ──────────── = 1
miles cancel → meters remain
meters cancel → miles remain
Orient each factor so an unwanted unit appears once above and once below. SI is the reference; customary units work when explicitly converted.
Takeaway: multiply by 1, arranged to leave the requested unit.
Worked chain · convert 60 mi/h to m/s
1 mi = 1609.344 m | 1 h = 3600 s
60 mi/h
× (1609.344 m / 1 mi) × (1 h / 3600 s)
mi cancels mi; h cancels h. The remaining unit is m/s, the requested speed unit.
60 × 1609.344 / 3600 = 26.8224 → 26.8 m/s
Numerical trail: 60 mi/h × (1609.344 m/mi) × (1 h/3600 s). The equal-to-one ratios change labels, not the speed itself; retain digits during the chain, then round the display as appropriate.
Read the leftover units before accepting the number.
Area · square the entire length factor
ft × ft
A square has two length directions: ft · ft = ft². Convert each direction, so the factor is squared too.
1 ft = 0.3048 m (exact)
10 ft² × (0.3048 m / 1 ft)²
= 10 ft² × (0.3048 m/ft)(0.3048 m/ft)
ft² cancels ft × ft below; m × m remains.
= 0.9290304 m²
Exact factor arithmetic shown; round the displayed result to suit the input’s precision. Unit conversion does not add measurement precision.
Volume · cube the whole factor
ft × ft × ft
Three length directions make ft³: each gets its own factor. Sketch indicates three dimensions; not to scale.
1 ft³ × (0.3048 m / 1 ft)³ = 0.028316846592 m³
Three ft factors cancel ft³; the three meter factors leave m³. In general: convert length by k, area by k², volume by k³.
Why area squares and volume cubes the factor
One length conversion for each independent direction
ft → m
A
B
(0.3048 m/ft)²
ft² × (m²/ft²) → m² Two ft cancel; two m remain.
C
D
(0.3048 m/ft)³
ft³ × (m³/ft³) → m³ Three ft cancel; three m remain.
E
Final check: the leftover unit must match the requested quantity. Use factors for compatible definitions/systems; conversion relabels the same quantity, not its precision.
Unit ledger · cancel the dimensions, then calculate
60 mi
1609.344 m
×
h
=
26.8224 m/s
mi
3600 s
The mi and h cross out because each occurs in a numerator and denominator. The only units left are m/s; the unrounded value is 26.8224 m/s, about 26.8 m/s.
Choose the factor orientation by the units you need to remove—not by the size of the fraction.
Pair check: mi ÷ mi = 1 · h ÷ h = 1; only m/s survives.
Orientation rule: put the unit to remove on the opposite side of the factor.
Compatible definitions are essential; the ratio equals one only for the stated units.
Takeaway · multiply by a ratio equal to 1, carry units through every step, and stop when the requested unit remains. Conversion changes the label—not the physical quantity or its measurement precision.
What cancels in the speed chain?
÷ mi = 1
÷ h = 1
Each matching unit in numerator and denominator contributes a factor of 1; the remaining unit trail is m/s.
Factor power = number of length directions
AREA
10 ft² × 0.3048² = 0.9290304 m² (exact-factor arithmetic)
VOLUME
Two ft cancel for m²; three ft cancel for m³. Keep the unit trail through each multiplication.
From [convert]: changing units must not pretend a measurement is more precise.
Measured reading vs exact count
From [check]: matching dimensions narrows candidate forms; it does not prove a law.
θ
One swing repeats after period T
Pendulum: propose a power law
T = C Lᵃ gᵇ
T = period (time); L = pendulum length; g = gravitational acceleration. Unknown exponents a, b are what dimensions can constrain; C is a dimensionless constant.
Write known dimensions in SI base dimensions: [T] = time; [L] = length; [g] = length/time².
This is a candidate model under stated variables and a power-law assumption—not a claim that units alone reveal the whole physical law.
Exponent ledger · match the length and time powers
factor / quantity length ℓ time τ
T (period) 0 1
C (dimensionless) 0 0
Lᵃ a 0
gᵇ b −2b
RHS: Lᵃgᵇ a + b −2b
Since [T] = time¹: a + b = 0; −2b = 1 → b = −½, a = ½
T = C √(L/g)
Check: [L/g] = length ÷ (length/time²) = time², so √(L/g) has units of time.
3. Numerical scale ≠ full prediction
L = 1.0 m; g = 9.8 m/s²
√(L/g) = √(1.0/9.8) s ≈ 0.32 s
This is the timescale from dimensions with C left unknown—not yet a complete period prediction.
Small-angle ideal pendulum: C = 2π → T ≈ 2π(0.32 s) ≈ 2.0 s
Dimensions fix a = ½ and b = −½, but cannot calculate the dimensionless constant C; the ideal small-angle physics supplies 2π.
4. What the inference assumes—and misses
Assume the listed variables are exhaustive (T depends only on L and g), and a power law T = C Lᵃgᵇ is a suitable model. Small-angle idealization is used only to compare C = 2π.
Omitted variables (such as swing amplitude outside the small-angle limit), dimensionless constants/groups, or an unsuitable alternative functional form can escape this simple exponent-matching method. Dimensions constrain a candidate; experiments or a physical model decide more.
Takeaway · dimensions narrow a model to T ∝ √(L/g); they do not uniquely determine the law or its constant.
What the simple inference can—and cannot—see
length L
assume
C is dimensionless
C unknown
gravity g
T = C√(L/g)
Assume these inputs are exhaustive; use a power law.
Needs physics or experiment; dimensions alone cannot set it.
If swing amplitude θ matters (e.g., beyond small angles), it is an omitted variable: the simple model may no longer describe the period. Other functional forms can also escape this power-law test.
Pipeline · from inference to real-world debug
pivot
From [infer]: pressure-test a candidate against the actual quantities, units, and interfaces.
apply the checks in practice
Five-pass diagnostic routine
1 Name the quantity: mass, force, area, temperature…
2 Put units beside every value—keep number + unit together.
3 Separate quantity types (especially mass from force).
4 Audit each operation and conversion; track units through it.
5 Check the final unit and whether the magnitude is plausible.
Mismatch? Investigate; unit checks catch many errors but cannot identify every bad assumption, input, or physical model.
1 · Mass is not weight (force)
mass · 10 lbm
weight · force
W = mg; g₀ = 32.174 ft/s² → W ≈ (10 lbm)(32.174 ft/s²)/g_c = 10 lbf
g_c = 32.174 lbm·ft/(lbf·s²) is the customary-unit conversion constant. A slug is another mass unit: 1 slug = 1 lbf·s²/ft. “lb” alone is ambiguous.
10 lbm ≈ 10 lbf weight at g₀—not interchangeable labels. Outside standard gravity, the weight changes; the mass does not.
2 · Area: the conversion factor is squared
CORRECT · 10 ft² × (0.3048 m / ft)²
ft² means ft × ft: convert both length directions.
= 10 × 0.3048² m² = 0.9290304 m² ≈ 0.929 m²
ft² cancels ft²; m² remains, matching area.
WRONG · × 0.3048 once gives 3.048 m·ft—not m²
The leftover unit reveals the missed second factor.
3 · Keep the unit trail through the operation
m = 3 kg; a = 2 m/s²; distance d = 2 m
F = ma = (3 kg)(2 m/s²) = 6 kg·m/s² = 6 N
E_work = Fd = (6 N)(2 m) = 12 N·m = 12 J
If units vanish: “12” can be mislabeled 12 N; trail shows it is 12 J (energy), not force.
4 · Temperature: interval vs absolute value
INTERVAL ΔT = 5 °C = 5 K same-size temperature change
ABSOLUTE T 20 °C = 293.15 K add 273.15 before T-ratios
Use kelvin in relations involving absolute temperature (e.g., ideal-gas law); Celsius is fine for a temperature difference.
5 · Interface check: Mars Climate Orbiter
thruster data lbf·s
receiver expects N·s
units not reconciled
1 lbf·s ≈ 4.44822 N·s · same impulse, different numerical scale
Prevention: declare units in the interface; convert explicitly; reject mismatched unit metadata before use.
The lbf·s/N·s interface mismatch was a major contributing error in the loss—not a claim that units alone explain every failure.
Prevent the mismatch at the interface
value + unit + quantity type
attach
interface contract expected unit
Compatible units → convert explicitly then check result unit + scale
Missing / incompatible units → stop No silent guess or reinterpretation
Units catch many interface and arithmetic errors—not wrong assumptions, bad inputs, or a flawed physical model.
Unit trail · the operation determines the result type
F = 6 N
d = 2 m
E = 12 J
N·m = (kg·m/s²)·m = kg·m²/s² = J
The arithmetic gives 12; the unit trail says work/energy. Calling it “12 N” would change the quantity to force, even though the bare number looks unchanged.
Final check: requested quantity? surviving unit? plausible scale?
Debug in order · verify before trusting
Quantity? → Units on every value? → Type and operation compatible?
Conversion done? → Requested final unit? → Plausible magnitude?
Keep the unit attached to the value: it records quantity type and scale.
Which SI base-unit expansion is correct for force?
A proposed distance equation is x = at. If [a] = L/T² and [t] = T, what is the dimension of the right-hand side—and what does that imply?
A speed of 60 mi/h is converted using 1 mi = 1609.344 m and 1 h = 3600 s. Which factor arrangement leaves m/s?
Convert 10 ft² to m² using 1 ft = 0.3048 m. Which setup correctly converts both length directions?
From [debug]: unit-aware checks protect the final engineering interface.
Review route · the result of each step feeds the next
Meaning tags the value → language names its dimensions → checking tests the model → conversion aligns units → precision limits claims → inference narrows formulas → debugging catches use and interface errors.
Using the board’s multiplication rule, how should 2.34 × 1.2 = 2.808 be reported?
Dimensional analysis gives a pendulum power-law form T = C√(L/g). What can dimensions alone determine?
For a relation that uses absolute temperature, what should 20 °C be converted to before using it?
A receiver expects impulse in N·s, but an interface supplies a value in lbf·s without converting it. What is the best unit-aware response?
validate
match? convert / reject
MEASURED · instrument-limited
Digits express the recorded resolution; the final digit is limited by the measurement.
Only measured inputs constrain precision; exact counts do not set a significant-figure limit.
Count significant figures: start at the first nonzero digit
Leading zeros only locate the decimal; zeros between significant digits count; trailing zeros count when the decimal/notation makes them explicit.
0.0042
m/s
1002
1.500
Scientific notation removes ambiguity: 1.5 × 10³ has 2 sig figs; 1.500 × 10³ has 4. Bare “1500” is ambiguous—state precision or use scientific notation.
Two rounding rules · different questions
2.34 × 1.2 = 2.808
Multiply · relative precision
1.2 has 2 significant figures (fewest), so report 2 significant figures.
2.808 → 2.8
Keep 2.808; after 2 sig figs the next digit is 0, so report 2.8.
Rule: × or ÷ → fewest sig figs
Why not one rule? Products scale relative precision; sums inherit the coarsest place value.
2.808 → 2.8 · retain working digits, round for reporting
Two significant figures keep 2 and 8; later digits guard the rounding decision.
.
8
ones
tenths