Engineering Foundations

LESSON 1.1

Units and Dimensional Reasoning

Make quantities comparable before calculating or making a design decision.

Learning objectives

• Convert length, time, and compound units with visible cancellation.
• Check an equation by comparing dimensions on both sides.
• Distinguish a numerical conversion from a change in the physical quantity.

Quantities carry meaning

• A number needs a quantity and a unit. “The length is 25” is incomplete because 25 millimeters and 25 meters describe very different objects. Record the symbol, value, and unit together: L = 25 mm. The unit is part of the information, not decoration added after calculation. Engineering calculations become easier to audit when every intermediate line preserves it.
• Convert by multiplying by one. One meter equals 1,000 millimeters, so the ratio 1 m / 1,000 mm equals one. Multiplying a quantity by this ratio changes its written form without changing the physical length. Choose the ratio orientation so unwanted units cancel. This approach also exposes mistakes in compound units, where memorized conversion shortcuts can hide the direction of the conversion.

Worked calculation

• Convert before multiplying. A fictional rectangular panel is 300 mm long and 0.20 m wide. First convert its length: 300 mm × (1 m / 1,000 mm) = 0.300 m. Then calculate area: A = 0.300 m × 0.20 m = 0.060 m². Multiplying 300 by 0.20 and calling the result 60 m² would mix units and overstate the area by a factor of 1,000.
• Square and cube conversion factors. Area uses two length factors. Therefore 1 m² = 1,000,000 mm², not 1,000 mm². The panel area can be checked independently as 300 mm × 200 mm = 60,000 mm²; dividing by 1,000,000 mm²/m² gives 0.060 m² again. Agreement between two calculation paths is useful evidence, although both paths still depend on the same supplied dimensions.

Check the model

• Use dimensions as an error detector. Speed has dimension length/time, so distance = speed × time is dimensionally consistent. Adding speed to time is not. Dimensional consistency does not prove that the correct physical law was chosen: distance = 2 × speed × time also has consistent dimensions. Use the check to reject impossible forms, then test the assumptions and numerical result separately.
• State precision and scope. The dimensions here are fictional inputs for a flat rectangle. The calculation ignores holes, curvature, and measurement uncertainty. Writing extra decimal places does not make the panel dimensions more certain. Preserve sufficient intermediate digits and report a sensible final precision, with the original inputs available for review.

Try, explain, and verify

• Independent exercise. A model vehicle travels 1.2 km in 4 minutes. Convert both inputs to SI units, calculate its average speed, and explain whether the result says anything about its maximum speed. Show the canceled units rather than presenting only a calculator result.
• Model reasoning. Distance is 1,200 m and elapsed time is 240 s, so average speed = 1,200 m / 240 s = 5 m/s. The same result is 18 km/h. Average speed summarizes the whole interval; stops or faster segments could occur within it. The common mistake is converting kilometers but leaving minutes unchanged while labeling the result m/s.

Applied scenario

A teammate submits two panel-area estimates that differ by a factor of 1,000. You must identify the unit mismatch before comparing material use.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

A 300 mm by 0.20 m rectangle has what area?

1. 60 m²
2. 0.060 m²
3. 0.000060 m²
4. 600 m²

LESSON 1.2

Measurement, Resolution, and Uncertainty

Separate what a measurement shows from what it can justify.

Learning objectives

• Distinguish instrument resolution, repeatability, and systematic error.
• Calculate a mean and a simple observed range.
• Report a result with a justified uncertainty statement.

Measurements are evidence with limits

• Resolution limits what can be read. A ruler marked every millimeter does not directly resolve hundredths of a millimeter. Some instruments support estimating between marks, but the method and uncertainty must be stated. An electronic display with many digits also needs a specification or calibration history; the display alone does not prove accuracy. Record the measurement method alongside the number.
• Repeatability and accuracy answer different questions. Repeatability describes agreement among repeated readings under stated conditions. Accuracy concerns agreement with an appropriate reference value. A misaligned ruler can produce tightly clustered readings that are all offset. Repeating the measurement can reveal variation, but it cannot automatically remove a shared bias such as starting from a damaged ruler edge.

Worked measurement record

• Summarize the observed readings. Four fictional length measurements are 100, 101, 100, and 99 mm. Their mean is (100 + 101 + 100 + 99) / 4 = 100 mm. The minimum is 99 mm and maximum is 101 mm. Report the observed range as 99–101 mm rather than claiming that every future reading must lie inside it.
• Describe the uncertainty method honestly. Half the observed range is 1 mm, so “mean 100 mm; half-range 1 mm across four readings” is a transparent descriptive summary. It is not automatically a statistical confidence interval or the instrument’s total uncertainty. Resolution, alignment, temperature, and reference calibration may add other effects. For this introductory exercise, keep those limitations visible instead of attaching an unsupported confidence percentage.

Improve the observation

• Control the measurement procedure. Define the endpoints, view the scale consistently, and repeat the same placement method. A change in procedure may change what quantity is being measured. When comparing two groups of readings, record whether the instrument or observer changed. Otherwise apparent improvement could reflect a different method rather than a better design.
• Propagate a simple bounded uncertainty. Suppose a rectangle has nominal dimensions 100 ± 1 mm and 50 ± 1 mm, where these are assumed hard bounds for this exercise. Its smallest modeled area is 99 × 49 = 4,851 mm²; its largest is 101 × 51 = 5,151 mm². The nominal area is 5,000 mm². This interval calculation assumes both dimension bounds can occur together and does not establish real manufacturing acceptance limits.

Try, explain, and verify

• Independent exercise. Readings from a second fictional ruler are 104, 104, 105, and 103 mm for the same reference whose stated length is 100 mm. Calculate the mean, inspect the spread, and explain why averaging more readings may not solve the difference from the reference.
• Model reasoning. The mean is 104 mm and observed range is 103–105 mm. The small spread suggests repeatability under this procedure, while the 4 mm mean difference suggests a possible systematic issue or an incorrect reference assumption. Check alignment, zero position, and reference documentation before calling the ruler accurate. The misconception is that a stable answer must be the right answer.

Applied scenario

A design team compares two measurement methods. One produces more consistent readings but is offset from the documented reference.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which conclusion follows from readings tightly clustered around 104 mm when a suitable reference is 100 mm?

1. The instrument is certainly accurate
2. Averaging guarantees a mean of 100 mm
3. The method may be repeatable but systematically offset
4. The reference must be wrong

LESSON 1.3

Engineering Requirements and Model Boundaries

Turn an open-ended idea into a testable learning problem.

Learning objectives

• Write measurable requirements and distinguish them from design choices.
• Identify inputs, outputs, constraints, and assumptions.
• Create an acceptance test that can reject an inadequate result.

Start with the need

• A requirement describes success. “Use a large wheel” names a possible solution. “Move the model 0.60 m on a flat diagram within 3 s” describes an outcome that different designs could satisfy. Specify the quantity, threshold, conditions, and method of checking it. Requirements that cannot be tested make it hard to tell whether a polished drawing actually addresses the original need.
• Constraints and preferences play different roles. A constraint sets a boundary such as a maximum modeled mass or a permitted material list. A preference helps rank designs that already satisfy the essential requirements. If a design violates a mandatory constraint, a high aesthetic score should not silently rescue it. Record who set the requirement and what evidence would justify changing it.

Worked problem definition

• Describe a fictional delivery cart model. The paper model must represent a 0.60 m journey completed in at most 3 s with a modeled payload of 0.50 kg. Its footprint must fit inside a 0.20 m by 0.15 m rectangle. These values are teaching assumptions, not instructions for building a powered cart. The expected output is a calculation sheet and annotated concept drawing, not an assertion of real-world performance.
• Derive a testable target. For the stated journey, average speed must be at least 0.60 m / 3 s = 0.20 m/s. That does not determine acceleration, wheel torque, or stopping distance. A constant-speed model can support an initial timing calculation, but a real design would need additional startup, braking, traction, and safety analysis. Keep each conclusion tied to the model that supports it.

Make the reasoning inspectable

• Draw the system boundary. State whether the model includes only the cart or also its controller, power source, route, and environment. Inputs might be payload mass and travel distance; outputs might be predicted time and footprint. An omitted effect is not necessarily unimportant. Listing omitted effects tells the next reviewer where the current evidence stops and what another model or experiment would need to add.
• Build acceptance tests before selecting a design. A reviewer should be able to inspect the drawing scale, reproduce the time calculation, and check the payload value. Include a failing case: a predicted time of 3.4 s does not meet the 3 s requirement even if the concept is visually attractive. A test is stronger when the pass criterion is set before seeing the result.

Try, explain, and verify

• Independent exercise. A proposal says, “Our cart is fast, lightweight, and safe.” Rewrite two statements as measurable requirements for the paper study, then identify one claim that cannot be demonstrated through timing arithmetic alone. Include conditions rather than choosing an unexplained number.
• Model reasoning. One valid timing requirement is “the constant-speed model predicts completion of 0.60 m within 3 s.” A valid size requirement is “the scaled drawing fits the specified footprint.” Safety cannot be established by those two checks; it needs identified hazards, appropriate evidence, and competent review for any physical application. The common error is treating a model prediction as a completed physical verification.

Applied scenario

A fictional client asks for a small delivery cart. You must produce a bounded paper study with traceable acceptance checks.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which statement is directly testable in the bounded cart study?

1. The cart is excellent
2. The cart is modern
3. The cart is universally safe
4. The predicted time for 0.60 m is at most 3 s

LESSON 2.1

Rearranging Equations and Estimating Results

Solve for an unknown while preserving physical meaning.

Learning objectives

• Rearrange a relationship using operations applied to both sides.
• Estimate an answer before using a calculator.
• Substitute the result back into the original equation.

Equations express a relationship

• Identify the unknown before manipulating symbols. An equation is not a recipe tied to one arrangement. For a constant-speed model, d = v × t relates distance, speed, and elapsed time. If distance and time are known, speed is the unknown; if speed is known, the same relationship can predict time. List known quantities with units and state why constant speed is an acceptable simplification for the exercise.
• Preserve equality through balanced operations. Divide both sides of d = v × t by t to obtain v = d / t, provided t is nonzero. Dividing by zero has no defined result here. The rearrangement should be visible so a reviewer can distinguish an algebra mistake from incorrect inputs. Memorizing multiple versions without understanding the transformation makes unfamiliar equations harder to handle.

Worked calculation

• Solve, label, and substitute. A fictional motion trace covers 2.4 m in 6.0 s at a modeled constant speed. The equation gives v = 2.4 m / 6.0 s = 0.40 m/s. Substitute the result back: 0.40 m/s × 6.0 s = 2.4 m. The recovered distance supports the algebra and arithmetic; it does not prove the trace actually had constant speed between its endpoints.
• Estimate the scale first. Since 2.4 is less than 6, the numerical result in m/s should be less than one. A calculator output of 4.0 m/s should immediately trigger a check. A rough estimate is not a substitute for the calculation, but it catches misplaced decimal points and inverted ratios. Explain what order of magnitude you expected and why.

Extend the method

• Handle an offset explicitly. A simplified calibration relationship is y = 2x + 3, where x and y are dimensionless display values. If y = 11, subtract 3 from both sides to get 8 = 2x, then divide by 2 to obtain x = 4. Dividing 11 by 2 first ignores the offset. Substitute 4 into the original relationship to recover 11.
• Check whether the relationship applies. A linear calibration may be valid only within a stated range, such as x from 0 to 5. A computed x = 8 would lie beyond the range of evidence even if the algebra were correct. This is the difference between solving an equation and justifying its use. Keep range limitations attached to the result when communicating it.

Try, explain, and verify

• Independent exercise. A fictional sensor conversion is T = 5r + 10, with T in degrees Celsius and r a dimensionless reading. Find r when T = 35°C. Then explain whether this equation alone establishes that a real sensor is calibrated accurately.
• Model reasoning. Subtract 10°C and divide by the slope 5°C per unit of r: r = 5. Substitution gives T = 5 × 5 + 10 = 35°C. The relationship needs calibration evidence and a valid operating range before use with a physical sensor. The misconception is that a numerically consistent formula proves the underlying instrument is accurate.

Applied scenario

A spreadsheet predicts travel times but one cell inverts distance and speed. You must find the algebra and unit error.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

For y = 2x + 3 and y = 11, what is x?

1. 4
2. 5.5
3. 7
4. 22

LESSON 2.2

Vectors, Components, and Direction

Represent direction explicitly instead of adding magnitudes blindly.

Learning objectives

• Distinguish scalars from vectors.
• Resolve a planar vector into perpendicular components.
• Add component vectors and check a resultant.

Direction changes the result

• A vector includes magnitude and direction. Distance traveled is a scalar, while displacement describes the change from starting position to ending position. Walking 3 m east and 3 m west gives 6 m of traveled distance but zero displacement. Choose a coordinate system and label positive directions before calculating. An unexplained minus sign often indicates that the reference direction was never made explicit.
• Components are projections, not extra vectors added to the original. In a two-dimensional plane, a vector can be represented by an x component and a perpendicular y component. Their vector sum equals the original vector. The component magnitudes usually do not sum to the original magnitude because the axes are perpendicular. A scale drawing can make that distinction visible before introducing trigonometric equations.

Worked vector calculation

• Use a right triangle with a stated angle. A 10 N force acts 30° above the positive horizontal axis. Its components are Fx = 10 cos(30°) ≈ 8.66 N and Fy = 10 sin(30°) = 5.00 N. The angle is measured from the x axis; measuring from the vertical would change which component uses sine. Keep the angular convention attached to the drawing.
• Reconstruct to verify. The resultant magnitude is sqrt(8.66² + 5.00²) ≈ 10.0 N, with the slight difference caused by rounding. The ratio Fy/Fx ≈ 0.577 corresponds to 30° in this quadrant. This calculation assumes perpendicular axes and a planar vector. It does not describe out-of-plane components or determine how a particular object responds to the force.

Combine several contributions

• Add along matching axes. Suppose a second force is 4 N directly west, so its components are -4 N and 0 N. The combined components are 4.66 N horizontally and 5.00 N vertically. The magnitude is sqrt(4.66² + 5.00²) ≈ 6.84 N. Simply adding 10 + 4 = 14 N ignores the directions and gives the wrong resultant for this configuration.
• Interpret signs physically. A negative x component means west relative to the chosen east-positive axis; it does not mean a force has a negative magnitude. If a calculation predicts an unexpected sign, inspect the coordinate choice and inputs before discarding it. A consistent coordinate convention lets another reader reproduce the result even if they would have chosen different positive directions.

Try, explain, and verify

• Independent exercise. A fictional position change is 3 m east followed by 4 m north. Draw its components, calculate the displacement magnitude, and distinguish that value from the total path length. Give the direction as an angle measured north of east.
• Model reasoning. The displacement magnitude is sqrt(3² + 4²) = 5 m, while the path length is 7 m. Its direction is arctan(4/3) ≈ 53.1° north of east. The common misconception is treating a component path and its diagonal resultant as the same traveled route. Both quantities are useful, but they answer different questions.

Applied scenario

A paper robot route contains eastward and northward segments. You must report both route length and start-to-finish displacement.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

A route goes 3 m east and 4 m north. What is the displacement magnitude?

1. 7 m
2. 1 m
3. 5 m
4. 12 m

LESSON 2.3

Free-Body Diagrams and Balanced Forces

Translate an object and its surroundings into a usable force model.

Learning objectives

• Choose a system boundary for a free-body diagram.
• Distinguish action-reaction pairs from forces on one object.
• Calculate a support reaction for a simple static case.

Isolate one system

• A free-body diagram shows external interactions. Draw the selected object alone, then show forces exerted on it by gravity, supports, contacts, and any applied loads. Do not include every visible arrow from the surrounding scene. A useful diagram states what the object is, where the forces act, and which positive directions are used. This translation is often more important than the arithmetic that follows.
• Mass and weight are related but different. Mass is measured in kilograms; weight is a gravitational force measured in newtons. Near Earth’s surface, a simplified calculation uses W = m × g with g = 9.81 m/s². A 2.0 kg object therefore has modeled weight 19.62 N. This course uses constant g for introductory examples and does not treat kilograms as a force unit.

Worked equilibrium calculation

• Check vertical balance. A fictional 2.0 kg block rests on a horizontal table with no other vertical force. Draw its downward weight W and upward normal force N. For zero vertical acceleration, sum Fy = N - W = 0, giving N = 19.62 N. The force balance supports this result only under the stated stationary, horizontal-contact assumptions.
• Change the loading, not the rule. Add a 5.0 N downward applied force to the same paper model. The balance becomes N - 19.62 - 5.0 = 0, so N = 24.62 N. The normal force is not automatically equal to weight. It adjusts to satisfy the contact and motion conditions in this simplified model. With acceleration or loss of contact, the assumptions would need revision.

Avoid misleading diagrams

• Newton’s third-law partners act on different bodies. The table pushes upward on the block, and the block pushes downward on the table. Those two forces are a third-law pair, but only the upward table-on-block force belongs on the block’s free-body diagram. The block’s weight and normal force can balance numerically without being a third-law pair, because both act on the block and arise from different interactions.
• Force balance alone is not the whole of equilibrium. A body can have zero net force yet experience an unbalanced turning effect. For a rigid body in planar static equilibrium, both the force sums and the moment sum must vanish. This lesson uses a centered block so rotation is not part of the calculation. A later lever or beam model must include distances from a pivot and torque balance.

Try, explain, and verify

• Independent exercise. In a stationary paper model, a 1.0 kg block on a table is pulled upward with a 3.0 N force while remaining in contact. Calculate its normal force. Identify the assumption that would fail if the upward pull exceeded the block’s weight.
• Model reasoning. Weight is 9.81 N. Vertical balance gives N + 3.0 - 9.81 = 0, so N = 6.81 N upward. An ordinary table cannot supply a downward normal contact force to hold the block in place. A computed negative normal would indicate that the assumed contact configuration is invalid, rather than a valid negative support magnitude.

Applied scenario

A teammate labels every support force as mg. You must explain why an added downward load changes the reaction.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

A stationary 2 kg block has a 5 N additional downward force. With g = 9.81 m/s², what normal force is predicted?

1. 14.62 N
2. 19.62 N
3. 5 N
4. 24.62 N

LESSON 3.1

Work, Energy, and Power

Track energy transfers and distinguish their amount from their rate.

Learning objectives

• Calculate work for a constant force parallel to displacement.
• Distinguish energy, power, and force through their units.
• Apply an energy balance with an explicitly stated efficiency.

Track the transfer

• Work connects force with displacement. For a constant force acting in the same direction as a straight displacement, W = F × d. The unit is a joule: 1 J = 1 N·m. More generally the force component along the motion matters. A perpendicular force does no work in this ideal translational calculation even if its magnitude is large. State the direction assumption before using the simplified equation.
• Power is a rate, not another word for energy. Average power is energy transferred divided by elapsed time: P = E / t. Its unit is the watt, with 1 W = 1 J/s. Two models can transfer the same energy with different power because they take different amounts of time. Avoid confusing a motor’s power rating with the total energy supplied during a particular interval.

Worked transfer calculation

• Calculate useful work and average power. A fictional constant force of 12 N moves a model load 0.50 m along the force direction. Work transferred to that motion is 12 × 0.50 = 6.0 J. If the transfer takes 2.0 s, the average useful power is 6.0 / 2.0 = 3.0 W. The calculation does not establish the instantaneous power during startup or any unmodeled acceleration.
• Include a stated efficiency. Suppose the useful work is 60% of the input energy in this bounded model. Efficiency η = Euseful / Einput, so Einput = 6.0 J / 0.60 = 10 J. The remaining 4 J must be accounted for in other transfers or stored energy within an appropriate boundary. It does not disappear. Treating 60% as the fraction lost would reverse the calculation.

Define the energy boundary

• Storage and dissipation must be distinguished. A moving body can store kinetic energy; a raised body can gain gravitational potential energy; warming can increase internal energy. “Loss” often means energy no longer available for the intended purpose, not destruction of energy. A diagram with input, useful output, and other transfers makes this distinction clearer than an unexplained efficiency percentage.
• Use ideal models for comparison, not certification. A simplified energy budget omits details such as time-varying friction, control electronics, and measurement uncertainty. It can show whether a claim violates the stated budget or which factor dominates. It cannot by itself verify a real actuator, battery, lifting mechanism, or safe operating limit. Keep the paper exercise separate from approval of physical equipment.

Try, explain, and verify

• Independent exercise. A fictional device receives 20 J and produces 15 J of useful output over 5 s. Calculate efficiency, useful average power, and the amount of input not counted as useful output. Explain why that amount should not be described as destroyed energy.
• Model reasoning. Efficiency is 15/20 = 0.75, or 75%. Useful average power is 15 J / 5 s = 3 W. The remaining 5 J goes to other transfers or storage under a complete energy balance. The misconception is that a higher efficiency necessarily means a higher power: a highly efficient slow transfer can have lower power than a less efficient rapid transfer.

Applied scenario

A paper actuator comparison reports one model in watts and another in joules. You need a consistent energy and timing comparison.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

A device receives 20 J and produces 15 J useful output. What is its efficiency?

1. 25%
2. 75%
3. 133%
4. 3%

LESSON 3.2

Graphs, Slopes, and Rates of Change

Extract physical meaning from a plotted relationship.

Learning objectives

• Calculate slope with units from a position-time graph.
• Distinguish an interval average from an instantaneous quantity.
• Recognize that a graph shape reflects a model and sampling choices.

Read the axes first

• A graph is a relationship between quantities. Before describing a line as steep or flat, read both axis labels and units. A position-time graph and a speed-time graph can have the same visual shape while representing different physical behavior. State the independent variable, dependent variable, and range. A cropped vertical scale can exaggerate changes without altering the numerical data.
• Slope is change in output divided by change in input. For position x against time t, the interval slope is Δx/Δt, with units m/s. It represents average velocity over that interval. If the line is straight, the model has constant velocity there. The numerical steepness also depends on plotting scales, so compare computed slopes rather than judging images with different axes by eye.

Worked graph interpretation

• Calculate an interval rate. A fictional trace gives x = 0.20 m at t = 1.0 s and x = 1.00 m at t = 5.0 s. The change in position is 0.80 m over 4.0 s, so average velocity is 0.20 m/s. Using x/t at the final point would give 0.20 m/s here by coincidence only if the chosen origin matched the relationship. Always use differences when finding slope between specified points.
• Inspect a nonzero intercept. Consider x = 0.50 + 0.20t, with x in meters and t in seconds. At t = 5 s, x = 1.50 m, but velocity remains 0.20 m/s. Dividing final position by time gives 0.30 m/s and incorrectly includes the initial 0.50 m offset. The intercept represents position at time zero, not an additional speed.

Look beyond one number

• Changing slopes reveal changing motion. Suppose positions at 0, 1, and 2 s are 0, 0.20, and 0.80 m. The first interval’s average velocity is 0.20 m/s and the second’s is 0.60 m/s. This suggests faster motion during the later interval. Three samples do not uniquely determine the full continuous path or exact instantaneous velocity between samples; many curves could pass through the same points.
• Connect rates with accumulation carefully. On a speed-time graph, the area under the curve represents distance traveled when speed is nonnegative. A constant 0.20 m/s for 4 s corresponds to a rectangle of area 0.80 m. On a velocity-time graph, signed area gives displacement. Negative segments can cancel positive displacement even though the total traveled distance remains positive.

Try, explain, and verify

• Independent exercise. A straight position-time line passes through (2 s, 1 m) and (6 s, 3 m). Calculate its slope, then determine the position at t = 0 under the straight-line model. Explain what new evidence would be needed to claim that the real motion was straight-line in time throughout the interval.
• Model reasoning. Slope is (3 - 1)/(6 - 2) = 0.50 m/s. The intercept is 1 - 0.50 × 2 = 0 m. Additional measurements and an appropriate motion model would support the behavior between samples. The common mistake is assuming that joining sparse points with straight segments proves what happened between them.

Applied scenario

A route dashboard plots position against time. A reviewer confuses its nonzero starting position with increased speed.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

For x = 0.50 + 0.20t in meters and seconds, what is the modeled constant velocity?

1. 0.50 m/s
2. 0.70 m/s
3. 0.20 m/s
4. It changes with time

LESSON 3.3

Numerical Models and Step-Size Checks

Use a small computational model without hiding its approximation.

Learning objectives

• Construct an explicit step-by-step numerical update.
• Compare numerical output with a known simple solution.
• Use a smaller step size to investigate approximation error.

Build a model you can inspect

• A numerical model updates a state using rules. A state might be position, speed, temperature, or stored energy. Specify its initial value, the update rule, the time step, and the number of steps. A spreadsheet can make each row inspectable. An output chart alone is insufficient because it hides the rule and inputs that produced the apparent behavior.
• Different errors need different checks. A coding or formula error means the implementation does not follow the intended rule. Approximation error can remain even when the rule is implemented correctly. Model-form error arises when the chosen relationship omits relevant physics. Comparing against an exact result for a simplified case helps check implementation and approximation, but does not validate every real application of the model.

Worked numerical trace

• Use a constant-acceleration example. Start with x = 0 m and v = 0 m/s, use a = 2 m/s², and choose Δt = 1 s. In an explicit Euler update, calculate xnew = xold + vold × Δt and vnew = vold + a × Δt. Both right-hand sides use old values. After one step, x = 0 m and v = 2 m/s; after two, x = 2 m and v = 4 m/s.
• Compare with the analytic reference. For these assumptions, exact position is x = 0.5at² = 4 m at t = 2 s, while exact velocity is at = 4 m/s. The numerical velocity matches, but the position is underestimated because each step used the beginning-of-step velocity while speed increased during the interval. Updating velocity first and using it for position creates a different algorithm, not the same explicit Euler method.

Test convergence and scope

• Repeat with a smaller step. With Δt = 0.5 s, the old velocities for four steps are 0, 1, 2, and 3 m/s. The position increments are 0, 0.5, 1.0, and 1.5 m, totaling 3.0 m at 2 s. This is closer to the 4 m reference than the 1 s-step result. Reduced error in this example illustrates step-size sensitivity; it does not guarantee that every smaller step solves all numerical problems.
• Preserve the assumptions. Acceleration is constant, the motion is one-dimensional, and no forces are calculated from real hardware. A longer or more complex model may need a different integration method, stability analysis, and validation data. State the modeled time range and compare at the same final time when changing step size. Otherwise different durations can masquerade as numerical convergence.

Try, explain, and verify

• Independent exercise. Repeat the explicit Euler position calculation with Δt = 0.25 s through t = 2 s. Predict whether the result should fall above or below 4 m before adding the increments. Keep old and new states in separate columns.
• Model reasoning. The eight old velocities are 0, 0.5, 1, 1.5, 2, 2.5, 3, and 3.5 m/s. Multiplying their sum, 14 m/s, by 0.25 s gives 3.5 m. It remains below the analytic position and is closer than 3.0 m. The misconception is that matching a velocity result guarantees that every modeled quantity is equally accurate.

Applied scenario

A spreadsheet animation looks smooth, but you must determine whether its position estimate changes when the time step changes.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Explicit Euler uses old velocity for position. For a = 2 m/s², x0 = v0 = 0, and two 1 s steps, what is x?

1. 2 m
2. 4 m
3. 6 m
4. 8 m

LESSON 4.1

Technical Sketches and Scaled Communication

Make a drawing that another learner can interpret and check.

Learning objectives

• Use a declared drawing scale and consistent units.
• Annotate dimensions independently of visual appearance.
• Separate concept sketches from verified fabrication drawings.

Draw to communicate a decision

• A sketch answers a question. A concept sketch may show overall arrangement, motion, or a force path without defining every manufactured feature. State its purpose so the reader knows what can be inferred. A drawing that looks realistic is not automatically precise enough to build from. Label views, significant interfaces, and any assumed symmetry instead of relying on appearance alone.
• Scale links paper length to model length. A scale of 1:5 means one unit on the drawing represents five of the same units in the modeled object. Convert units before comparing. The written dimensions normally describe the object, not the length measured on a printed or resized image. A screenshot can change apparent scale while leaving the intended dimensions unchanged.

Worked scale calculation

• Lay out a fictional cart footprint. The concept footprint is 200 mm long and 150 mm wide. At 1:5, draw a rectangle 40 mm by 30 mm on paper. Label its object dimensions as 200 mm and 150 mm and write “concept plan view, 1:5 when printed at specified size.” If the page is resized, the dimension labels remain the primary source for the intended model sizes.
• Add an interface with clear reference points. Place a modeled axle centerline 30 mm from the object’s front edge. Its drawing offset is 30/5 = 6 mm. State that the dimension is measured from the front edge to the centerline, not to the wheel’s outer surface. Choosing explicit reference features prevents two reasonable readers from interpreting the same number differently.

Include limits and consistency checks

• Multiple views should agree. A plan view can show length and width but cannot alone establish height. Add a side view for a modeled height of 80 mm, drawn as 16 mm at 1:5. Align relevant features or label their locations. Check that the same 200 mm length appears consistently across views; contradictory dimensions require resolution rather than averaging.
• Distinguish nominal dimensions from tolerances. “200 mm” states a nominal size but does not define an allowable manufacturing variation. A tolerance must be justified by fit, function, process, and inspection needs. This introductory concept exercise does not prescribe real tolerances, material strength, fasteners, or safe operating limits. Mark unverified details as assumptions so a concept sketch is not mistaken for a released engineering drawing.

Try, explain, and verify

• Independent exercise. A second paper study uses a 1:4 scale for the same 200 mm by 150 mm footprint and an 80 mm height. Determine the three drawing dimensions and explain why a measured screenshot width should not override a written object dimension.
• Model reasoning. Drawing length is 50 mm, width 37.5 mm, and height 20 mm. Screenshots and printers may resize the image, so measuring the displayed rectangle can give a misleading object size. The misconception is that a detailed illustration is automatically a production drawing. Complete requirements, interfaces, tolerances, material choices, and appropriate review would still be needed for a physical design.

Applied scenario

A learner shares a resized screenshot of a model cart drawing. You must retain the intended dimensions without trusting the displayed scale.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

At 1:5 scale, how long is a 200 mm object drawn on paper?

1. 1,000 mm
2. 40 mm
3. 200 mm
4. 5 mm

LESSON 4.2

Design Tradeoffs and Evidence-Based Selection

Compare feasible options using transparent criteria and assumptions.

Learning objectives

• Separate mandatory constraints from scored preferences.
• Calculate a weighted comparison with declared score meanings.
• Test whether a recommendation depends strongly on chosen weights.

Decide what the comparison means

• Filter feasibility before scoring preferences. An attractive option cannot compensate for failing an essential requirement unless that requirement is formally changed. Begin with pass/fail constraints such as the required footprint or modeled travel time. Only compare preference scores among options that pass. Keep missing evidence distinct from failure: an untested requirement is unresolved, not automatically satisfied.
• Scores need a defined direction and basis. If a high score is better, say so and define what a score represents for each criterion. Cost scores, for example, may be normalized estimates rather than actual currency. A table of unexplained numbers creates an illusion of precision. Preserve the source or reasoning for each score so the recommendation can be challenged constructively.

Worked comparison

• Calculate a simple weighted score. Two fictional feasible concepts are rated from 1 to 5, with 5 better. Concept A scores 5 for low cost and 2 for serviceability. Concept B scores 3 for low cost and 5 for serviceability. With equal weights of 0.5, A scores 0.5 × 5 + 0.5 × 2 = 3.5; B scores 0.5 × 3 + 0.5 × 5 = 4.0. Under those preferences, B ranks higher.
• Test a changed preference. If low cost receives weight 0.8 and serviceability 0.2, A scores 4.4 while B scores 3.4. The ranking reverses. This is useful information, not a calculation failure: the recommendation depends on stakeholder priorities. Explain why the selected weights fit the actual decision rather than adjusting them after seeing which concept wins.

Keep uncertainty visible

• A weighted sum makes assumptions. It treats the scored criteria as compensable and assumes the score increments are meaningful enough to combine. Those assumptions may be weak for subjective ratings. Use the table as a structured conversation and sensitivity check, not an automatic authority. A major safety or compliance condition should be a mandatory gate with appropriate review, not an easily outweighed preference score.
• Gather evidence where it can change the decision. If serviceability scores are guesses, inspect what task, tools, access, and time would support a rating. Additional detail has greatest value when it resolves uncertainty that could alter feasibility or ranking. Record the next verification action and an owner rather than presenting a disputed score as settled fact.

Try, explain, and verify

• Independent exercise. Let the low-cost weight be w and serviceability weight be 1 - w. Write the score equations for A and B and solve for the weight at which their scores are equal. Then explain which side of that threshold favors each concept.
• Model reasoning. A = 5w + 2(1 - w) = 2 + 3w. B = 3w + 5(1 - w) = 5 - 2w. Equality gives 5w = 3, so w = 0.60. Below 0.60, B ranks higher; above it, A ranks higher. At 0.60 both score 3.8. The common error is reporting a single winner without explaining the priorities and evidence that made it win.

Applied scenario

A fictional design review must choose between two feasible cart concepts, but the owner and maintenance reviewer value different criteria.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

At equal low-cost and serviceability weights, A has scores 5 and 2 while B has 3 and 5. Which weighted result is correct?

1. A = 7, B = 8
2. A = 4.0, B = 3.5
3. A = 3.5, B = 4.0
4. Both = 5.0

LESSON 4.3

Engineering Foundations Capstone: A Reviewable Paper Study

Combine requirements, mathematics, sketches, and evidence into one coherent design argument.

Learning objectives

• Assemble a traceable requirements-to-evidence portfolio.
• Check units, calculations, and model limitations across a small study.
• Defend a recommendation and identify the next validation step.

Build one connected argument

• The capstone is a paper study, not a product approval. Evaluate a fictional cart concept that must represent a 0.60 m journey within 3 s, carry a modeled 0.50 kg payload, and fit within a 0.20 m by 0.15 m footprint. Your deliverable combines calculations, a scaled drawing, an option comparison, and a verification plan. Explain what each artifact supports and which claims remain outside the model.
• Create a traceability table. Give each requirement an identifier, a stated condition, and a named evidence item. For example, R1 is the time target and E1 is the constant-speed calculation. R2 is the footprint and E2 is the dimensioned plan view. Evidence can fail to satisfy a requirement; traceability means the relationship is inspectable, not that every row must show a pass.

Worked baseline study

• Check the timing calculation. A proposed constant-speed model uses 0.25 m/s, so predicted time is 0.60 m / 0.25 m/s = 2.4 s. This passes the bounded 3 s timing requirement with a 0.6 s numerical margin. It does not include startup, stopping, obstacles, or slip. Label that margin as a model result rather than a guaranteed reserve in physical operation.
• Check size and payload consistently. A 1:5 concept drawing of the maximum footprint measures 40 mm by 30 mm on correctly sized paper, while its dimension labels state 200 mm by 150 mm. The modeled payload weight is 0.50 kg × 9.81 m/s² = 4.905 N. Payload weight alone does not determine wheel loading, structure strength, or actuator sizing; the cart’s own mass and force geometry are not supplied.

Compare, challenge, and communicate

• Use a failing case to test the review process. A second timing proposal uses 0.15 m/s and predicts 4.0 s, so it fails R1 under the same assumptions. A professional review should preserve that finding rather than rounding it into a pass. Comparing the two cases shows that the acceptance check can discriminate between proposals, although neither case validates the constant-speed assumption itself.
• State the next evidence needed. A physical development program would require appropriate supervision, hazard review, validated motion behavior, material and structural analysis, and relevant design controls. For this course, identify the first missing evidence item and explain how it could change the recommendation. Keep the proposed next step at the planning level; do not infer that the paper study authorizes construction or powered testing.

Try, explain, and verify

• Independent exercise. A reviewer changes the journey to 0.90 m but keeps the 3 s limit. Recalculate the original 0.25 m/s proposal, find the minimum constant speed needed, and update the traceability table. Identify which artifacts require revision and which footprint information can remain unchanged.
• Model reasoning. The original proposal now predicts 0.90/0.25 = 3.6 s and fails. Minimum average speed under the constant-speed assumption is 0.90/3 = 0.30 m/s. The timing calculation and recommendation need revision; the stated footprint can remain if no size requirement changed. The misconception is that a previously approved result remains valid after its input requirements change.

Applied scenario

A reviewer asks you to defend a fictional cart concept and then changes one requirement during the paper-study review.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

A 0.25 m/s constant-speed proposal must travel 0.90 m within 3 s. What does the model predict?

1. 2.4 s, pass
2. 3.6 s, fail
3. 0.75 s, pass
4. 3.0 s, exactly pass

LESSON 5.1

Bound a measurement before comparing designs — worked model

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Learning objectives

• Bound a measurement before comparing designs
• 4,900.5 to 5,100.5 mm²
• Defend the conclusion with scoped and reproducible evidence.

The practice brief

• A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Understand the mechanism

• The nominal area is 100 × 50 = 5,000 mm², or 0.005 m². For positive lengths, the minimum product occurs when both dimensions are at their minima: 99 × 49.5 = 4,900.5 mm². The maximum is 101 × 50.5 = 5,100.5 mm².

• These bounds are slightly asymmetric around the nominal product because multiplication includes a cross term. A first-order estimate adds the relative bounds, 1/100 + 0.5/50 = 2%, giving approximately ±100 mm², while the exact interval is −99.5 to +100.5 mm². Keep the approximation and exact method separately labeled.

• A bound on an input is not automatically a confidence interval; you have not been given a distribution, coverage probability, or correlation model. The stated rectangular geometry also matters. Holes, curvature, thermal changes, and a biased instrument are outside this fixture.

• Changing mm² into m² divides every area by 1,000,000; it does not reduce uncertainty. Report the original dimensions, conversions, nominal calculation, limiting combinations, and excluded effects together so another learner can reproduce the argument. A narrower displayed decimal is not a better measurement. Compare designs only when both use a compatible measurement rule and an explicit acceptance requirement.

Build the initial evidence table

• Write one row for each supplied input or observation. Keep identifiers, units, timestamps, roles, counts, and states distinct where the case uses them. Label information supplied by the fixture separately from a conclusion you calculate. Do not add an unstated business rule to make the answer easier.

Worked reasoning

4,900.5 to 5,100.5 mm²

Check the scope

• Explain what this evidence supports and what it leaves unresolved. A correct calculation, a parsed object, a matching identifier, and an authorized action are different kinds of claims. Use only the claims established by this specific case.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

What is the original exact area interval?

1. 4,900.5 to 5,100.5 mm²
2. Exactly 5,000 mm² with no uncertainty
3. 4,999.5 to 5,000.5 mm²
4. 0.0049 to 0.0051 mm²

LESSON 5.2

Bound a measurement before comparing designs — controlled variation

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Learning objectives

• Bound a measurement before comparing designs
• Only the width bound tightens to ±0.1 mm; length remains 100 ±1 mm. The new interval is 99 × 49.9 = 4,940.1 to 101 × 50.1 = 5,060.1 mm². The narrower width bound reduces the area range, but the original length uncertainty remains.
• Defend the conclusion with scoped and reproducible evidence.

Keep the original baseline

• A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

The controlled change

• Only the width bound tightens to ±0.1 mm; length remains 100 ±1 mm. The new interval is 99 × 49.9 = 4,940.1 to 101 × 50.1 = 5,060.1 mm². The narrower width bound reduces the area range, but the original length uncertainty remains.

Compare both paths

• Retain the original result and change only the stated condition. Predict the new result first, then trace the affected relationship, policy decision, state transition, or denominator. Mark the earliest step where the old and new reasoning differ. A paper trace is a legitimate exercise, but it is not an executed test.

Reference reasoning

• The nominal area is 100 × 50 = 5,000 mm², or 0.005 m². For positive lengths, the minimum product occurs when both dimensions are at their minima: 99 × 49.5 = 4,900.5 mm². The maximum is 101 × 50.5 = 5,100.5 mm².

• These bounds are slightly asymmetric around the nominal product because multiplication includes a cross term. A first-order estimate adds the relative bounds, 1/100 + 0.5/50 = 2%, giving approximately ±100 mm², while the exact interval is −99.5 to +100.5 mm². Keep the approximation and exact method separately labeled.

• A bound on an input is not automatically a confidence interval; you have not been given a distribution, coverage probability, or correlation model. The stated rectangular geometry also matters. Holes, curvature, thermal changes, and a biased instrument are outside this fixture.

• Changing mm² into m² divides every area by 1,000,000; it does not reduce uncertainty. Report the original dimensions, conversions, nominal calculation, limiting combinations, and excluded effects together so another learner can reproduce the argument. A narrower displayed decimal is not a better measurement. Compare designs only when both use a compatible measurement rule and an explicit acceptance requirement.

Review the changed conclusion

• Only the width bound tightens to ±0.1 mm; length remains 100 ±1 mm. The new interval is 99 × 49.9 = 4,940.1 to 101 × 50.1 = 5,060.1 mm². The narrower width bound reduces the area range, but the original length uncertainty remains.

Investigate disagreement

• If your result differs, inspect the first diverging step. Check whether you changed the population, units, time window, authority, or outcome definition. Record the correction rather than hiding the failed attempt.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which conclusion follows from the changed practice case?

1. The changed condition cannot affect the result.
2. The original result proves every future outcome.
3. The evidence gap can be filled with an assumption.
4. Only the width bound tightens to ±0.1 mm; length remains 100 ±1 mm. The new interval is 99 × 49.9 = 4,940.1 to 101 × 50.1 = 5,060.1 mm². The narrower width bound reduces the area range, but the original length uncertainty remains.

LESSON 5.3

Bound a measurement before comparing designs — evidence and review

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Learning objectives

• Bound a measurement before comparing designs
• Prepare a nominal-and-bound table, show both unit systems, and compare exact bounds with the first-order estimate. State whether a claimed exact area of 5,000.000 mm² is justified.
• Defend the conclusion with scoped and reproducible evidence.

Your deliverable

Prepare a nominal-and-bound table, show both unit systems, and compare exact bounds with the first-order estimate. State whether a claimed exact area of 5,000.000 mm² is justified.

Inputs and constraints

• A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Reference mechanism

• The nominal area is 100 × 50 = 5,000 mm², or 0.005 m². For positive lengths, the minimum product occurs when both dimensions are at their minima: 99 × 49.5 = 4,900.5 mm². The maximum is 101 × 50.5 = 5,100.5 mm².

• These bounds are slightly asymmetric around the nominal product because multiplication includes a cross term. A first-order estimate adds the relative bounds, 1/100 + 0.5/50 = 2%, giving approximately ±100 mm², while the exact interval is −99.5 to +100.5 mm². Keep the approximation and exact method separately labeled.

• A bound on an input is not automatically a confidence interval; you have not been given a distribution, coverage probability, or correlation model. The stated rectangular geometry also matters. Holes, curvature, thermal changes, and a biased instrument are outside this fixture.

• Changing mm² into m² divides every area by 1,000,000; it does not reduce uncertainty. Report the original dimensions, conversions, nominal calculation, limiting combinations, and excluded effects together so another learner can reproduce the argument. A narrower displayed decimal is not a better measurement. Compare designs only when both use a compatible measurement rule and an explicit acceptance requirement.

Acceptance review

4,900.5 to 5,100.5 mm²

Required variation

• Only the width bound tightens to ±0.1 mm; length remains 100 ±1 mm. The new interval is 99 × 49.9 = 4,940.1 to 101 × 50.1 = 5,060.1 mm². The narrower width bound reduces the area range, but the original length uncertainty remains.

Review with another engineer

Explain the problem, the selected method, your actual evidence, a plausible alternative, and the remaining limit. A reviewer should be able to reconstruct the case from the retained inputs and steps without relying on a polished screenshot.

Portfolio evidence

• Keep a concise README with the fictional fixture, assumptions, original result, changed-input trace, any executed test command and output, and one limitation. Clearly label designs and predictions that were not executed. These examples do not authorize production changes or prove production readiness.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A fictional flat panel has length 100 mm with a stated bound of ±1 mm and width 50 mm with a bound of ±0.5 mm. Treat the bounds as independent allowable ranges, not as statistical standard deviations. No physical cutting or material purchase is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which deliverable addresses this case?

1. A universal success claim without tests.
2. A copy of a result without a method.
3. Prepare a nominal-and-bound table, show both unit systems, and compare exact bounds with the first-order estimate. State whether a claimed exact area of 5,000.000 mm² is justified.
4. A screenshot without its input or scope.

LESSON 6.1

Check a numerical model by refining the step — worked model

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Learning objectives

• Check a numerical model by refining the step
• The h=0.25 result, 0.31640625
• Defend the conclusion with scoped and reproducible evidence.

The practice brief

• A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Understand the mechanism

• With step h=0.5, two updates give 1 → 0.5 → 0.25. With h=0.25, four updates give 1 → 0.75 → 0.5625 → 0.421875 → 0.31640625. The supplied exact reference is exp(−1) ≈ 0.36787944.

• Absolute errors are approximately 0.11787944 and 0.05147319. The smaller-step result is closer for this particular rule, interval, and method, but agreement between two approximations is not a general proof of correctness. Inspect the update equation as well as the output.

• A program using 1+h instead of 1−h would model growth, and running more steps would not repair that sign error. Record time and value in separate columns, and verify that the final number of steps reaches the same final time in both runs. Do not compare two steps at h=0.25 with two at h=0.5 and call the difference a resolution effect: those runs end at different times.

• Keep units and the model boundary explicit if adapting this arithmetic pattern to a physical problem later. This fixture is a deterministic numerical exercise, not a measured cooling curve. A model could converge numerically and still fail to represent the intended physical process.

Build the initial evidence table

• Write one row for each supplied input or observation. Keep identifiers, units, timestamps, roles, counts, and states distinct where the case uses them. Label information supplied by the fixture separately from a conclusion you calculate. Do not add an unstated business rule to make the answer easier.

Worked reasoning

The h=0.25 result, 0.31640625

Check the scope

• Explain what this evidence supports and what it leaves unresolved. A correct calculation, a parsed object, a matching identifier, and an authorized action are different kinds of claims. Use only the claims established by this specific case.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which approximation is closer to exp(−1)?

1. The h=0.5 result, 0.25
2. Both are exact
3. The run ending at a different final time
4. The h=0.25 result, 0.31640625

LESSON 6.2

Check a numerical model by refining the step — controlled variation

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Learning objectives

• Check a numerical model by refining the step
• Set h=1.5 and take one update. The result is −0.5 at time 1.5, although the supplied exact decay solution remains positive. This coarse-step result exposes a numerical limitation; it does not show that a real positive quantity must become negative.
• Defend the conclusion with scoped and reproducible evidence.

Keep the original baseline

• A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

The controlled change

• Set h=1.5 and take one update. The result is −0.5 at time 1.5, although the supplied exact decay solution remains positive. This coarse-step result exposes a numerical limitation; it does not show that a real positive quantity must become negative.

Compare both paths

• Retain the original result and change only the stated condition. Predict the new result first, then trace the affected relationship, policy decision, state transition, or denominator. Mark the earliest step where the old and new reasoning differ. A paper trace is a legitimate exercise, but it is not an executed test.

Reference reasoning

• With step h=0.5, two updates give 1 → 0.5 → 0.25. With h=0.25, four updates give 1 → 0.75 → 0.5625 → 0.421875 → 0.31640625. The supplied exact reference is exp(−1) ≈ 0.36787944.

• Absolute errors are approximately 0.11787944 and 0.05147319. The smaller-step result is closer for this particular rule, interval, and method, but agreement between two approximations is not a general proof of correctness. Inspect the update equation as well as the output.

• A program using 1+h instead of 1−h would model growth, and running more steps would not repair that sign error. Record time and value in separate columns, and verify that the final number of steps reaches the same final time in both runs. Do not compare two steps at h=0.25 with two at h=0.5 and call the difference a resolution effect: those runs end at different times.

• Keep units and the model boundary explicit if adapting this arithmetic pattern to a physical problem later. This fixture is a deterministic numerical exercise, not a measured cooling curve. A model could converge numerically and still fail to represent the intended physical process.

Review the changed conclusion

• Set h=1.5 and take one update. The result is −0.5 at time 1.5, although the supplied exact decay solution remains positive. This coarse-step result exposes a numerical limitation; it does not show that a real positive quantity must become negative.

Investigate disagreement

• If your result differs, inspect the first diverging step. Check whether you changed the population, units, time window, authority, or outcome definition. Record the correction rather than hiding the failed attempt.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which conclusion follows from the changed practice case?

1. The original result proves every future outcome.
2. The evidence gap can be filled with an assumption.
3. Set h=1.5 and take one update. The result is −0.5 at time 1.5, although the supplied exact decay solution remains positive. This coarse-step result exposes a numerical limitation; it does not show that a real positive quantity must become negative.
4. The changed condition cannot affect the result.

LESSON 6.3

Check a numerical model by refining the step — evidence and review

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Learning objectives

• Check a numerical model by refining the step
• Build both time/value tables, compute errors against the supplied reference, and diagnose the coarse-step negative value. Keep equation correctness, step sensitivity, and physical validity as separate checks.
• Defend the conclusion with scoped and reproducible evidence.

Your deliverable

Build both time/value tables, compute errors against the supplied reference, and diagnose the coarse-step negative value. Keep equation correctness, step sensitivity, and physical validity as separate checks.

Inputs and constraints

• A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Reference mechanism

• With step h=0.5, two updates give 1 → 0.5 → 0.25. With h=0.25, four updates give 1 → 0.75 → 0.5625 → 0.421875 → 0.31640625. The supplied exact reference is exp(−1) ≈ 0.36787944.

• Absolute errors are approximately 0.11787944 and 0.05147319. The smaller-step result is closer for this particular rule, interval, and method, but agreement between two approximations is not a general proof of correctness. Inspect the update equation as well as the output.

• A program using 1+h instead of 1−h would model growth, and running more steps would not repair that sign error. Record time and value in separate columns, and verify that the final number of steps reaches the same final time in both runs. Do not compare two steps at h=0.25 with two at h=0.5 and call the difference a resolution effect: those runs end at different times.

• Keep units and the model boundary explicit if adapting this arithmetic pattern to a physical problem later. This fixture is a deterministic numerical exercise, not a measured cooling curve. A model could converge numerically and still fail to represent the intended physical process.

Acceptance review

The h=0.25 result, 0.31640625

Required variation

• Set h=1.5 and take one update. The result is −0.5 at time 1.5, although the supplied exact decay solution remains positive. This coarse-step result exposes a numerical limitation; it does not show that a real positive quantity must become negative.

Review with another engineer

Explain the problem, the selected method, your actual evidence, a plausible alternative, and the remaining limit. A reviewer should be able to reconstruct the case from the retained inputs and steps without relying on a polished screenshot.

Portfolio evidence

• Keep a concise README with the fictional fixture, assumptions, original result, changed-input trace, any executed test command and output, and one limitation. Clearly label designs and predictions that were not executed. These examples do not authorize production changes or prove production readiness.

Offline practice

• [Download the BEAT IT engineering and quantum arithmetic checks](/resources/academy/engineering-quantum-checks.py). Run with Python 3 in a practice folder. The program uses only the standard library, makes no network calls, and checks synthetic arithmetic and ideal-model results.

• Passing these checks does not establish real equipment safety, hardware execution, or deployment readiness. Retain your predictions and the observed program output separately.

Further study

• [MIT OpenCourseWare: numerical methods](https://www.ocw.mit.edu/courses/18-03sc-differential-equations-fall-2011/pages/unit-i-first-order-differential-equations/numerical-methods/)

Applied scenario

A dimensionless teaching quantity follows the rate rule y' = −y with y(0)=1. Use the supplied forward-Euler update y_next = y_current × (1 − h). Compare approximations at time 1; no calculus derivation or physical experiment is required.

Your evidence and reasoning

Inputs and assumptions → diagram and equation → observed result → units and limits → one improvement.

Practice check

Which deliverable addresses this case?

1. A copy of a result without a method.
2. Build both time/value tables, compute errors against the supplied reference, and diagnose the coarse-step negative value. Keep equation correctness, step sensitivity, and physical validity as separate checks.
3. A screenshot without its input or scope.
4. A universal success claim without tests.