Industrial Engineering: Product Design, Work System Design and Facility Design

A good product design meets its function reliably, is safe and ergonomic to use, looks right (aesthetics), can be made and assembled economically with the processes available (producibility), uses materials and standard parts sensibly, can be maintained, and meets its cost target. Because most of a product's lifetime cost is committed by decisions taken at the design stage, long before much of it is spent, cost and quality are designed in rather than inspected in. The designer therefore balances performance against cost at every choice of material, tolerance and process.

Tolerance design decides how much variation each dimension may have. When parts stack in an assembly, the worst-case method adds tolerances linearly, T_assembly = ΣT_i, guaranteeing every assembly fits but forcing tight, costly part tolerances. The statistical (root-sum-square) method assumes independent, normally distributed, centred dimensions and adds them as T_assembly = √(ΣT_i²): parts of ±0.02, ±0.03 and ±0.06 mm give ±0.11 mm worst case but only ±0.07 mm statistically, so wider, cheaper part tolerances achieve the same assembly tolerance with a small, known risk. Taguchi's quality loss function L(y) = k(y − m)² puts a cost on any deviation from the target m, not only on out-of-specification parts: if a deviation of 0.5 mm costs Rs 100, k = 400 Rs/mm² and a 0.2 mm deviation costs Rs 16. Robust design then chooses parameter settings that make the product insensitive to noise.

• Standardisation fixes sizes, materials, methods and parts to agreed norms, so that parts interchange, inventories shrink, economies of scale grow and quality becomes repeatable; preferred numbers (geometric series) give a rational spread of sizes.
• Simplification reduces the number of varieties — models, sizes, grades — by eliminating the superfluous ones, cutting set-ups, tooling and stock.
• Diversification adds products or varieties to spread risk, use spare capacity or reach new markets — the opposite pull, which is why the three are studied together: a firm simplifies within a product line while diversifying across lines.

Value engineering (applied at design) and value analysis (applied to an existing product) seek the required function at the lowest total cost without sacrificing quality or reliability: value = function/cost. Economists distinguish use value (what it does), esteem value (what makes it desirable), cost value and exchange value. Functions are described by a verb and a noun — 'transmit torque', 'support load' — and classified as basic or secondary; the function analysis system technique (FAST) diagram orders them by asking how and why. The job plan runs: information (gather facts, costs), function analysis, creative or speculative (brainstorm alternatives), evaluation, development or recommendation, and implementation with follow-up.

Concurrent (simultaneous) engineering replaces the sequential, 'over-the-wall' hand-off from design to process planning to manufacturing with cross-functional teams — design, manufacturing, quality, purchasing, marketing and suppliers — working on the product and its processes in parallel from the start. Problems are found while they are cheap to fix, engineering changes after release fall, and time to market shortens. Design for X names the life-cycle concern the design is optimised for: DFM (manufacture — processes, tolerances and materials that are easy to make), DFA (assembly — fewer parts, self-locating features, top-down insertion, no fasteners where snap fits will do), DFMA combining the two, DFE (environment), DFR (reliability), design for serviceability, disassembly and recycling, and design for cost. The Boothroyd-Dewhurst DFA index = 3N_min/t_total, where N_min is the theoretical minimum number of parts (a part is essential only if it must move, must be of a different material, or must be separate for assembly or service) and 3 s is taken as an ideal handling-and-insertion time per part: 4 essential parts in a 60 s assembly give 3 × 4/60 = 20 %.

F. W. Taylor's scientific management replaced rule of thumb with systematic study: time study to set a fair day's work, standardised tools and methods, scientific selection and training of workers, a division of responsibility between management (planning) and workers (doing) — including functional foremanship, with separate foremen for each function — and a differential piece-rate that paid more per piece to workers who met the standard. Frank and Lillian Gilbreth founded motion study: they broke manual work into seventeen elemental motions called therbligs (search, select, grasp, transport loaded, position, assemble, use, release, inspect, hold, rest and so on — an eighteenth, find, was added later), filmed work with a clock in view for micro-motion study, recorded both hands against time on a SIMO chart, and studied fatigue. Lillian Gilbreth extended the work to the psychology of management.

Productivity is output divided by input. Partial productivity uses one input (labour productivity = units per labour-hour); total-factor productivity uses labour and capital; total productivity divides the value of all output by the value of all inputs — labour, materials, capital, energy and other expenses: output worth Rs 60 000 from inputs of Rs 10 000, 20 000, 5 000, 3 000 and 2 000 gives 60 000/40 000 = 1.5. A rise in productivity means more output from the same input or the same output from less — not simply more output, which extra input could buy.

Method study systematically records and critically examines existing and proposed ways of doing work to develop easier, more effective methods. Its steps: select the job, record the facts, examine them critically (asking what, where, when, who and how — and why each), develop the best method, define, install it and maintain it. Recording uses the outline process chart; the flow process chart, with five ASME symbols — operation (circle), inspection (square), transport (arrow), delay (D) and storage (triangle); the two-handed chart; the multiple-activity chart for worker-machine interaction; the flow diagram and the string diagram (a thread traced over a scale plan to measure distance travelled); and the travel chart. Micro-motion study analyses short, repetitive, high-volume tasks down to therbligs from film. The principles of motion economy fall in three groups: use of the human body (both hands begin and end together, move symmetrically and simultaneously, use the lowest body member that can do the job, prefer smooth curved motions to sharp changes of direction), arrangement of the workplace (fixed positions for tools and materials, within the normal working area, gravity-feed bins and drop delivery, suitable lighting and chair height), and design of tools and equipment (jigs, fixtures or foot pedals to free the hands, combined tools, pre-positioned tools).

Time study times a qualified worker with a stopwatch over enough cycles, element by element, and rates the observed pace against a concept of normal pace. Normal time = observed time × rating/100, and standard time = normal time × (1 + allowance), where allowances — personal needs, fatigue and unavoidable delay — are normally expressed as a percentage of the normal time. An element observed at 0.80 min with a rating of 110 % has normal time 0.88 min; with 15 % allowances the standard time is 0.88 × 1.15 = 1.012 min. The number of cycles to observe is set statistically from the scatter of the pilot readings and the confidence and accuracy wanted.

Work sampling (activity sampling) makes a large number of instantaneous observations at random times and estimates the proportion p of time spent in an activity — idle, working, waiting. Its size follows the binomial: for an absolute accuracy ±e, n = z²p(1 − p)/e²; for a relative accuracy ±s (a fraction of p), n = z²(1 − p)/(s²p). With p = 0.25, z = 2 for about 95 % confidence and s = 0.10, n = 4 × 0.75/(0.01 × 0.25) = 1200 observations. It needs no stopwatch, suits long and irregular work and group activities, and is cheap per observation. Standard data are element times compiled from previous studies, often as formulas of the variables (length, weight), to set standards for new jobs without timing them. Predetermined motion time systems assign times to basic motions from tables: MTM (methods-time measurement, in time measurement units, 1 TMU = 0.00001 h = 0.036 s), Work-Factor and MOST (Maynard operation sequence technique); they set standards before a job exists and need no performance rating, because the rating is built into the tables.

Ergonomics fits the work to the worker. Anthropometry supplies body dimensions by percentile: clearances (door height, leg room) are designed for the 95th percentile so the large can pass, reaches and control forces for the 5th percentile so the small can manage, and adjustable features (seat height) span the range. The work environment — illumination, noise (exposure in dB(A) and its duration), heat and humidity, vibration — and the design of displays and controls (compatibility of movement, coding by shape and colour) determine fatigue, error and injury. Job evaluation ranks jobs, not people, to build a fair wage structure: non-quantitative methods are ranking and grading (classification); quantitative methods are factor comparison and the point method, which scores each job on weighted factors such as skill, effort, responsibility and working conditions. Merit rating (performance appraisal) rates the person in the job: graphic rating scales, straight ranking, paired comparison (each employee compared with every other, N(N − 1)/2 pairs), checklists, forced choice, critical incidents and behaviourally anchored scales.

Location factors: proximity to markets and to raw materials, availability and cost of labour, transport links, utilities (power, water), land and construction cost, taxes, incentives and regulation, community and climate. Evaluation methods: the factor-rating method scores each site on weighted factors and sums them; the centre-of-gravity method places a distribution facility at X = Σx_iW_i/ΣW_i, Y = Σy_iW_i/ΣW_i, W the volume shipped to each point; the load-distance method picks the site minimising Σ(load × distance); locational break-even (cost-volume) analysis compares total cost F + vQ at each site and finds the volume ranges over which each is cheapest; the transportation model optimises shipping from several sources; and the Brown-Gibson model combines objective (cost) and subjective factor measures. For sites A (fixed Rs 2 lakh, variable Rs 10 per unit) and B (Rs 1 lakh, Rs 15), costs cross where 200 000 + 10Q = 100 000 + 15Q, at Q = 20 000: B is cheaper below, A above.

Evaluating a layout. Systematic layout planning (Muther) rates the closeness needed between every pair of departments on a relationship chart — A absolutely necessary, E especially important, I important, O ordinary, U unimportant, X undesirable — and builds a layout honouring them. Quantitatively, a layout is scored by the load-distance (flow-distance) total Σf_ij d_ij, or its cost equivalent. CRAFT is an improvement algorithm: starting from an existing layout it swaps pairs of departments whenever the swap lowers the flow-cost total. ALDEP and CORELAP are construction algorithms that build a layout from the relationship chart.

Line balancing assigns tasks to workstations, respecting precedence, so that no station's work exceeds the cycle time C = available production time/required output. The theoretical minimum number of stations is N_min = ⌈Σt_i/C⌉ (rounded up). Line efficiency = Σt_i/(N × C), balance delay = 1 − efficiency, and the smoothness index SI = √[Σ(S_max − S_i)²] measures how evenly the work is spread. Heuristics: the largest candidate rule (assign the longest task whose predecessors are done and that fits), the Kilbridge-Wester column method (group tasks into precedence columns), and the ranked positional weight method (rank tasks by their own time plus the times of all tasks that follow them).

Material handling adds cost but no value, so the aim is to move less, not to move faster. Its principles: plan the system as a whole; move material in unit loads (pallets, containers) rather than piece by piece; minimise distance and the number of handlings, keeping flow in straight lines; use gravity wherever possible; standardise and mechanise equipment; keep the handled load flexible to changes; and consider energy, safety and ergonomics. Equipment falls into conveyors (belt, roller, chain, overhead trolley — fixed path, high volume), cranes and hoists (EOT, jib and gantry cranes — heavy loads over an area), industrial trucks (forklifts, pallet trucks and automated guided vehicles — variable paths), automated storage and retrieval systems (AS/RS) with computer-controlled stacker cranes, and positioning and unitising equipment.