23-Ind-B2 Manufacturing Processes · December 2013
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
National Exams — December 2013 — 98-Ind-B2 Manufacturing Processes. Closed book; Casio or Sharp approved calculators only. Any five of the seven questions constitute a complete paper; all questions are of equal value (20 marks each). Answers are written in point form but fully, with all calculations shown, as instructed. Complete answers to all seven questions follow.
Reference texts: Groover, Fundamentals of Modern Manufacturing: Materials, Processes, and Systems, 6th ed. — material selection, casting, metal-cutting theory, welding processes, polymer processing, statistical process control; Montgomery, Introduction to Statistical Quality Control, 8th ed. — acceptance sampling, control charts, the Deming/Taguchi quality philosophies.
Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.
Statistical process control (SPC) is the use of statistical methods to monitor and control a process while it is running, so problems are detected and corrected before defective product is produced, rather than caught by inspecting finished output. Its core elements are: control charts (e.g. $\bar{X}$-R, $\bar{X}$-S charts for variables; p, np, c, u charts for attributes), which plot a sample statistic over time against statistically derived upper and lower control limits (typically $\pm3\sigma$ around the process mean) to distinguish common-cause (inherent, random) variation from special-cause (assignable) variation; process capability analysis ($C_p$, $C_{pk}$), which compares the natural process spread to the engineering specification limits to judge whether a process that is in statistical control is also capable of meeting the tolerance; sampling and measurement systems, since SPC depends on periodically measuring a representative subgroup of output; and a defined response/corrective-action procedure, so that when a chart signals an out-of-control condition (a point beyond the control limits, or a non-random run/trend pattern), the operator has a documented procedure to investigate and eliminate the assignable cause before continuing production.
Acceptance sampling is a statistical technique for deciding whether to accept or reject a lot of incoming or outgoing product by inspecting a randomly drawn sample from the lot rather than every unit (100% inspection), based on a sampling plan that specifies the sample size $n$ and an acceptance number $c$ (the maximum number of nonconforming units the sample may contain for the lot to still be accepted). It is used when 100% inspection is impractical, destructive, or too costly, and it accepts a defined statistical risk of wrongly accepting a bad lot or rejecting a good one in exchange for much lower inspection cost.
The acceptance quality level (AQL) is the maximum percent (or proportion) nonconforming that, for purposes of acceptance sampling, is considered satisfactory as a process average — it is the quality level the sampling plan is designed to accept with high probability, not a promise that every accepted lot is that good. A sampling plan built around a stated AQL (e.g. under ANSI/ASQ Z1.4 or the equivalent ISO 2859-1 tables) balances the producer's risk (a good lot, at or better than the AQL, being wrongly rejected) against the consumer's risk (a bad lot being wrongly accepted), summarized by the plan's operating-characteristic (OC) curve.
Deming's approach treats quality as a management-system responsibility, not an inspection function: his 14 Points and the "System of Profound Knowledge" call for continuous improvement of every process (the Plan-Do-Check-Act, PDCA, cycle), driving out fear so workers report problems honestly, breaking down barriers between departments, ending reliance on mass inspection in favour of building quality in at the source, and reducing variation using statistical methods — famously, Deming taught that roughly 85–94% of quality problems are caused by the system (management-controlled) rather than by individual workers, so quality improvement must start with management action on the process, not with blaming or exhorting the workforce.
Taguchi's approach is a design-stage (off-line) quality-engineering methodology built on three main ideas: the quality loss function, which models quality loss as increasing (quadratically, in his formulation) with any deviation from the target value — not just deviation beyond a tolerance limit — so a part exactly on target is worth more than one merely "within spec"; robust design (parameter design), using designed experiments (orthogonal arrays) to find factor settings that make product/process performance insensitive to uncontrollable "noise" factors (material variation, environmental conditions, wear) without necessarily eliminating the noise itself; and the signal-to-noise ratio as the optimization criterion, maximizing the ratio of desired performance to variability rather than simply maximizing or minimizing the mean response. Where Deming's methods focus on managing and continuously improving the production system, Taguchi's methods focus on designing products and processes at the outset so they are inherently robust to variation.
