🎓 CSSBB Quest › Section VII

Section VII — Improve Phase ~14%

Improve carries 21 of the 150 scored questions. The 2022 BoK groups it into design of experiments, lean methods, and implementation. Questions test the BB’s ability to find the settings that improve the process, remove waste, and select, pilot and implement solutions. The Improve tollgate is where the Champion decides whether to scale.

Improve subtopics

Subtopic	Key exam content
A. Design of experiments (DOE)	Terminology (factors, levels, response, treatment, replication, randomization, blocking); planning; one-factor experiments; two-level full and fractional factorial designs; main effects, interactions; resolution (III, IV, V)
Solution selection	Solution selection matrix (weighted criteria), Pugh matrix, PICK chart (payoff vs. difficulty), piloting logic
B. Lean methods	Waste elimination (pull, kanban, 5S, standard work, mistake-proofing); cycle-time reduction (continuous flow, single-piece flow, setup reduction/SMED); kaizen and kaizen blitz; future-state value stream maps
Waste analysis	Eight wastes (DOWNTIME: Defects, Overproduction, Waiting, Non-utilized talent, Transportation, Inventory, Motion, Extra-processing)
Mistake-proofing (poka-yoke)	Prevention vs. detection poka-yoke; levels (shutdown, control, warning); examples by mistake type
C. Implementation: piloting and scaling	Pilot plan, success criteria, risk assessment (FMEA revisited), stakeholder communication, change management
C. Implementation: planning and evaluation	Pilot tests and simulations, implementation plan, communication and training plan, evaluating results against the goal

DOE resolution guide

Resolution	What’s confounded	When to use
III	Main effects with 2-factor interactions	Screening (many factors, few runs) — not for optimization
IV	2-factor interactions with each other (main effects only with 3-factor interactions)	Characterization; main effects are clear
V	2-factor interactions with 3-factor interactions	Reliable estimates of main effects and 2-factor interactions
Full factorial	Nothing confounded	When all interactions must be estimated

A 2^k full factorial needs 2^k runs per replicate (e.g., 4 factors = 16 runs); a half fraction 2^k−1 needs half as many (8 runs).

Eight lean wastes (DOWNTIME)

Defects — rework, scrap, warranty claims
Overproduction — making more than the customer needs now
Waiting — idle time for people or machines
Non-utilized talent — skills, ideas, and improvement suggestions that go unused
Transportation — unnecessary movement of materials between steps
Inventory — excess work-in-process or finished goods
Motion — unnecessary movement of people (reaching, walking, searching)
Extra-processing — doing more work than the customer requires or the spec demands

Exam trap: In the Toyota Production System view, overproduction is the worst waste because it creates or hides the others (inventory, waiting, transport, defects).

Takt time

Takt time sets the pace of production to match customer demand:

Takt time = Available production time ÷ Customer demand rate

Example: 480 min/day available ÷ 120 units/day demand = 4 min/unit takt time

If cycle time > takt time → process can’t meet demand → bottleneck
If cycle time < takt time → step has spare capacity; running it flat out without pull risks overproduction

Mistake-proofing hierarchy

Level	Type	Example
Best	Elimination — redesign so error is impossible	Asymmetric connector that can only go in one way
	Prevention — physical/process feature stops the error	Jig that won’t close unless part is correctly oriented
	Detection — alerts when error occurs or is about to	Vision system that rejects a misassembled unit
Weakest	Warning — notifies operator; relies on human action	Light or alarm that operator must respond to