Question 7 of 8: Centralized Inspection, Computers in Quality Control, and the Educational Value of SQC

Question 7: Centralized Inspection, Computers in Quality Control, and the Educational Value of SQC (20 marks)

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.

Part (a) — where centralized inspection earns its keep. Inspection may be organized at the point of manufacture, with gauges and inspectors stationed at each machine, or centrally, with work brought to a dedicated inspection crib or laboratory. Centralized inspection is preferable whenever the measurement itself is demanding rather than the flow of work. The clearest case is precision small-part manufacture — watch, instrument and aerospace components — where the characteristic to be verified is measured in micrometres and the measurement requires an optical comparator, a coordinate measuring machine, a surface-finish tester or a hardness tester. Those instruments are expensive, they demand a controlled environment of stable temperature, low humidity and freedom from vibration and airborne swarf, and they require a trained metrologist rather than a machine operator. Duplicating them at every workstation would be unaffordable, and using them beside a running machine tool would destroy their accuracy. A second and equally strong case is destructive or laboratory testing: tensile and impact tests, chemical analysis, metallographic sections, radiography and pressure testing all belong in a central laboratory because of the equipment, the hazard and the certification required. A third case arises where independence matters — final acceptance of purchased lots, first-article inspection of a new part, or verification of a characteristic on which product safety and professional liability depend — because an inspector reporting through a central quality organization is insulated from the production pressure that acts on a machine-side checker. Centralization also gives better utilization of gauges and inspectors, easier calibration control, tidier record keeping and a natural home for the statistical analysis of the results. Its costs are real and must be acknowledged: material must be transported and queued, so feedback to the operator is slower and a process can drift for longer before it is caught, and work in process rises. For high-volume, low-precision, fast-cycle work the balance tips back the other way, toward operator self-inspection at the machine supported by control charts and periodic audit.

Part (b) — the value of data processing and computers in a quality programme. Computers contribute at every stage of the quality cycle. In data capture, electronic gauges, digital calipers, coordinate measuring machines and in-process sensors feed measurements directly into the record without transcription, eliminating the copying errors that once made hand-kept charts unreliable and allowing sampling rates that no clerk could sustain. In analysis, control limits, capability indices, histograms, Pareto diagrams and scatter plots are produced automatically, and analyses that are impractical by hand — regression of a defect rate against several process variables, gauge repeatability and reproducibility studies, designed experiments, cumulative-sum and exponentially weighted moving-average charts — become routine. In real-time control, the computer applies the run rules as each point is plotted, alarms the operator within seconds of a shift, and in an adaptive installation adjusts a tool offset automatically, which converts detection into prevention. In reporting and traceability, the system links each measurement to lot, machine, operator, tool and material batch, so that a field failure can be traced to its production window and a recall bounded; this is also the mechanism by which a firm demonstrates conformity to ISO 9001 or to a customer's supplier-quality requirement. In cost of quality accounting, scrap, rework, warranty and appraisal costs are collected from the same transactions that drive the production and financial systems, which is what allows management to see quality as money rather than as an opinion. Finally, in supplier management, vendor rating and incoming-lot histories are maintained automatically and shared electronically with the vendor, which is exactly the continuous check the vendor in Question 6 needed. The caution worth stating is that computerization multiplies the consequences of a bad measurement system or a wrong control rule: a poorly conceived programme simply produces wrong answers faster and in greater volume.

Part (c) — how statistical quality control promotes understanding and appreciation of quality control. Statistical quality control changes the conversation about quality from assertion to evidence, and that is its educational value. A control chart posted at the machine shows the operator, in a single picture and in the units of the job, what the process normally does and when it has genuinely changed. It therefore teaches the single most important idea in the subject — the distinction between common-cause variation, which is the process behaving as designed and must be improved by management action on the system, and assignable-cause variation, which is a specific event the operator can often find and remove. Once that distinction is understood, over-adjustment stops, and with it the tampering that used to double the variability of a stable process. The chart also gives the operator ownership: it is plotted at the machine by the person running it, it responds within minutes, and it makes the operator the first line of quality rather than the subject of someone else's inspection report. For supervisors and managers, capability indices and Pareto analyses turn quality into a quantity that can be budgeted, targeted and tracked alongside output and cost, and the demonstration that most defects come from a few causes redirects effort from exhortation to engineering. Between customer and vendor, a shared control chart replaces argument about a rejected lot with a common record of what the process was doing when the lot was made. And because the method is a sampling method, it teaches that inspection can never sort quality into a product: the same statistics that justify accepting a lot on a sample also show how many defectives a screening operation will miss, which is the argument that finally moves a plant from detection to prevention. That progression — from inspecting quality in, to controlling the process, to designing variation out — is learned far more convincingly from a plant's own charts than from any training course.