Build your preparation around model identification: for every practice problem, name the model and its assumptions before touching the calculator. Use the decision table, the two worked scenarios, and the cue-first drill to make that habit automatic, then follow the adaptive preparation sequence and the readiness checks.
Read the Assumptions Before You Compute
Industrial formulas are assumption-sensitive: demand rate, service discipline, server count, and who produces the item all change which equation applies. Train yourself to name the model from the scenario's wording before calculating anything.
Consider the difference between EOQ and EPQ. Both use annual demand, a fixed cost per order or setup, and a holding cost per unit, yet one assumes the full quantity arrives at once while the other assumes you produce it yourself at a finite rate. If you compute EOQ for an in-house production scenario, you understate the order quantity and misstate holding cost, and every downstream number inherits the error.
The same identification discipline applies across the whole specification. In queueing, a second server changes the entire waiting-time formula. In statistics, paired data call for a different test than two independent samples. In economics, unequal alternative lives call for annual worth rather than present worth. The practical exercise below converts this from advice into a trainable, measurable skill.
- Cue-first drill: take ten mixed practice problems from the free practice page. For each, before any calculation, write three lines: the model name, the handbook section you would open, and the exact cue words that triggered your choice.
- Then solve only the problems you identified correctly, and re-read the underlying distinction for each one you missed.
- Self-check rubric: 9-10 correct identifications means move on to timed sets; 6-8 means rebuild the distinction notes for your misses; 5 or fewer means redraw the decision table by hand from memory.
- Expected observation: identification errors cluster where two formulas share symbols, such as demand rate appearing in both inventory and queueing contexts, which tells you exactly which distinctions to study.
Engineering Economics: Equalize Lives Before Comparing
When two alternatives have different service lives, comparing their present worths directly is the trap. Annual worth, or a common-multiple horizon, restores a fair comparison because it normalizes cost to a per-year basis.
Worked scenario: Machine X costs $40,000 with $9,000 annual operating cost over a 4-year life; Machine Y costs $60,000 with $6,000 annual operating cost over 6 years, at 10% interest. The plausible mistake is comparing present worths directly: X gives about $68,529 and Y about $86,132, so X looks cheaper. But X must be replaced mid-horizon, so its 4-year present worth does not cover the same service period.
The better decision converts each to annual worth: multiply each present worth by its capital recovery factor. X: $68,529 times 0.31547 gives roughly $21,620 per year; Y: $86,132 times 0.22961 gives roughly $19,781 per year. Y is actually cheaper per year of service. Why it matters: the ranking reverses, so the shortcut does not merely round differently, it selects the wrong machine. Whenever lives differ, either annualize or extend to a common horizon before choosing.
Applied Probability and Statistics: Match the Distribution to the Story
Each named distribution models a specific data-generating story: counts of events per interval, successes in a fixed number of trials, or times between events. Learn those stories and the test choice follows from how the data were collected.
Trace the contrasts. A Poisson count applies when events occur independently at a constant average rate per interval, such as arrivals per hour. A binomial applies when you run a fixed number of independent trials and count successes with constant probability. The exponential describes time between Poisson events. If a scenario says defects per shift, that cues Poisson; if it says 200 parts inspected with a known defect probability, that cues binomial.
Hypothesis-testing choices follow the same logic of matching structure to method. Two independent samples call for a two-sample test; the same units measured before and after call for a paired procedure, which removes unit-to-unit variation. Attribute data, such as pass versus fail, lead to proportion methods rather than means-based ones. Practice by writing one sentence naming the data structure before selecting any formula, because the structure, not the arithmetic, is where the choice is actually made.
Inventory Models: EOQ, EPQ, and Quantity Discounts Differ by Sourcing
The sourcing story decides the model: purchased items arriving in full shipments use EOQ, items produced in-house at a finite rate use EPQ, and price breaks at order-size thresholds require checking whether the discount is feasible.
Worked scenario: annual demand is 2,000 units, setup cost is $100 per production run, holding cost is $2 per unit per year, and the plant produces at 10,000 units per year. The plausible mistake is applying the basic EOQ formula: the square root of 2 times 2,000 times 100 divided by 2 gives about 447 units. That treats the item as purchased and ignores that inventory builds gradually during production.
The better decision recognizes finite production and uses EPQ, which divides the EOQ by the square root of one minus the demand-to-production ratio. With a ratio of 2,000 over 10,000, that is 447 divided by the square root of 0.8, giving 500 units per run. Why it matters: because inventory accumulates while production runs, more can be ordered per cycle without raising peak holding cost, and the run size, cycle length, and cost figures all shift. Decide first whether the scenario purchases or produces, then choose the formula.
| Model | Cue in the scenario | Key assumption | The mistake to avoid |
|---|---|---|---|
| EOQ | Ordering a purchased item | Full quantity arrives at once | Using it for in-house production |
| EPQ | Producing the item yourself | Finite production rate, gradual build-up | Forgetting the demand-to-production ratio |
| Quantity discount | Price breaks at order thresholds | Feasibility check at each break point | Grabbing the lowest price without checking total cost |
| M/M/1 queue | One server, Poisson arrivals | Single channel, unlimited capacity | Applying it when a second server exists |
| M/M/s queue | Multiple parallel servers | Shared single queue, s channels | Using the single-server waiting formula |
Queueing and Little's Law: Check Servers and Capacity First
Queueing results depend on structure, not just arrival and service rates. Confirm the number of servers, whether capacity is finite, and the queue discipline before choosing a waiting-time formula.
Little's Law is the safe anchor: average number in the system equals arrival rate times average time in the system, and the same relation holds for the queue alone. It applies broadly, which makes it useful for checking consistency among given values. If a problem gives two of the three quantities, compute the third, then use the specialized model formulas for utilization and waiting time rather than guessing from intuition.
The identification step matters most here because adjacent models differ sharply. Adding a parallel server reduces waiting time nonlinearly, so plugging a two-server scenario into a single-server formula overstates congestion. A finite-capacity model blocks arrivals when full, which changes both the effective arrival rate and the state probabilities. When you read a queueing scenario, say aloud: how many channels, what capacity, what discipline. Then open the corresponding handbook equations.
Production Design and Facilities: Bottlenecks, Balance, and Flow
Line output is set by the bottleneck station, and layout quality is judged by flow. Compute cycle time from demand, check station workloads against it, and use flow-based tools for layout and location comparisons.
In line balancing, translate required output into a target cycle time, which is available production time divided by required output. Then assign tasks so no station's total task time exceeds that cycle time, and check efficiency by comparing summed station times against the number of stations times the cycle time. A station loaded above the cycle time caps the whole line regardless of how lightly the other stations are loaded, so identify the constraint before redistributing work.
Facilities problems reward the same flow-first reading. From-to charts quantify interaction between department pairs, so a layout evaluation means asking which pairings carry the largest flows or material-handling costs and whether the arrangement shortens them. Location comparison methods weigh distance, volume, and cost structures differently, so name the method the scenario implies, such as a centroid-type calculation versus a factor comparison, before computing. Draw the flow picture first; the numbers then have somewhere to land.
Human Factors, Safety, and Systems: Model the Person and the Failure
This area asks you to quantify human performance and system reliability rather than rely on inspection. Learn the standard time-and-error concepts and the series-parallel reliability structure, then apply them to short paper scenarios.
Human factors content centers on fitting tasks to people: reaction and response time, information load, workstation fit, and fatigue or environmental effects on performance. Build a paper-scenario safety exercise: for a described hazard, write down the likelihood and consequence pair, list two candidate controls, and rank them using the hierarchy that prefers eliminating the hazard or engineering it out over warnings and training. Practicing that written ranking trains the reasoning you need when two options both look reasonable.
Reliability in systems engineering is structural. Series components multiply their reliabilities, so any single failure fails the system, while parallel redundancy allows the system to survive a failure. Trace a given block diagram into that series-parallel skeleton before computing, and note when a standby unit behaves differently from active parallel operation. Readiness checks for this whole guide: derive EPQ from EOQ without notes; convert present worth to annual worth and back; state which model a one-sentence cue points to within about thirty seconds. These are learning milestones, not predictions of any score.
References and further reading
Use these references to explore the concepts and check the latest information from the relevant organizations.