Agribusiness Management · Functional core · also ABM 505, POM (IRMA)
Designing and running production: location, layout, planning, inventory, quality and productivity, applied to processing plants and agribusiness operations.
In the exams: agribusiness management PhD entrance and fellowship examinations, production and operations (heavy)
Production and operations management (POM) is the design, running and improvement of the system that turns inputs such as raw produce, labour, energy, packaging and capital into finished goods and services. In agribusiness the system is usually a processing plant: a rice or flour mill, a dairy, a sugar factory, an oilseed crushing unit, a fruit-pulp line, a seed-conditioning plant, a feed mill or a cold store.
Agri-processing adds two features that general textbooks treat lightly. Raw material is seasonal and perishable, so a plant often runs flat out for a harvest season and idles or switches products for the rest of the year. Raw material varies in quality from load to load, so grading, sampling and statistical control are daily work, not a quality-department afterthought.
This subject is examined most heavily in the Agri-Business Management paper of India's ICAR AICE-JRF/SRF (PhD) examination, where operations is a named unit, and in UGC-NET Management and ASRB NET at a similar level.
This manual keeps to the operations function inside the plant. Network design, transport and distribution belong to the supply-chain subject, and the mathematics of linear programming, queuing and the economic order quantity belong to operations research; they are mentioned here only where a plant manager needs them.
Every organisation has three core functions: marketing creates demand, finance provides capital, and operations produces the good or service. Operations strategy is the set of long-term decisions about the production system that supports the business strategy. Wickham Skinner argued in 1969 that manufacturing was the missing link in corporate strategy, because firms left plant decisions to engineers without asking what the plant had to be good at. His later idea of the focused factory holds that a plant doing a limited set of tasks well beats one trying to do everything.
The usual competitive priorities are cost, quality, speed or dependability of delivery, and flexibility (of product mix and volume). Terry Hill distinguished order qualifiers, which a firm must meet to be considered at all (for example, meeting a food-safety certificate a retailer requires), from order winners, which make the customer choose it over rivals (for example, a lower price or shorter lead time).
Capacity is the maximum rate of output of a facility, expressed in output units per period (tonnes of paddy per hour, litres of milk per day) or, where output is mixed, in input units such as machine-hours. Design capacity is the maximum output under ideal conditions. Effective capacity is the maximum achievable after allowing for planned losses such as product changeovers, scheduled maintenance, cleaning-in-place cycles, meal breaks and quality checks. Actual output is what the plant really produced, which is lower again because of unplanned losses such as breakdowns, shortage of raw material, absenteeism and rejects.
Two ratios follow. Utilisation equals actual output divided by design capacity. Efficiency equals actual output divided by effective capacity. Example: a parboiling rice mill is rated at 10 tonnes of paddy per hour; after cleaning and changeover losses its effective capacity is 8 tonnes per hour; it actually processes 6 tonnes per hour. Utilisation is 6/10 = 60 per cent and efficiency is 6/8 = 75 per cent. Students often reverse the denominators; remember that efficiency judges the plant against what was achievable, utilisation against the nameplate.
Capacity planning asks how much, when and where. Strategies are to lead demand (add capacity ahead of expected growth), lag demand (add only after demand is proven) or match it in smaller steps. Agri-processing capacity is complicated by seasonality: a sugar mill may crush only during the cane season, so its annual output depends on season length as well as hourly capacity. Bottleneck analysis says the capacity of a sequential process equals the capacity of its slowest stage.
Productivity is the ratio of output to input. It measures how well resources are used, whereas production measures only how much is made. A single-factor or partial productivity uses one input, such as litres of milk processed per labour-hour or tonnes of flour per kilowatt-hour. Multifactor productivity divides output by a combination of inputs, all converted to money. Total factor productivity in a plant setting includes all inputs: labour, materials, energy, capital and other expenses.
Example: a dairy plant processes 40,000 litres in a week with 50 workers on 40-hour weeks, giving 40,000/2,000 = 20 litres per labour-hour. Next week it processes 46,200 litres with 55 workers on 40-hour weeks, giving 46,200/2,200 = 21 litres per labour-hour, a 5 per cent rise in labour productivity even though output rose 15.5 per cent.
Partial measures can mislead: replacing workers with machines raises labour productivity while capital productivity may fall. Multifactor measures correct this.
Productivity rises through technology (better machines, automation), method improvement (work study, layout changes), people (training, incentives, participation) and management (planning, maintenance, quality). In agri-processing, yield (output of good product per unit of raw material, such as head rice recovery or oil extraction rate) is often the single most valuable productivity measure because raw material is the largest cost; the attainable yield varies with variety, moisture and season, so plants benchmark against their own history and similar plants rather than a universal figure.
Process choice is the decision about how production is organised. Five types are commonly taught, arranged from low volume and high variety to high volume and low variety.
A project produces a unique, one-off output with a defined start and finish, such as building a new cold store or commissioning a milk-processing plant; resources come to the site. A job shop (jobbing) makes small quantities of many different products, each to customer order, using general-purpose machines and skilled workers; a custom seed-treating or small engineering workshop that repairs processing equipment is an example. Batch production makes moderate quantities of a product, then changes over to another; many bakeries, spice grinders, pickle makers and fruit-processing units run in batches, because each product needs a different recipe and the equipment is cleaned between runs. Mass or line production makes large volumes of a standard product on a dedicated line in which each workstation does the same task repeatedly; a bottling or pouch-packing line is an example. Continuous-flow production processes a homogeneous material without interruption, often around the clock, with high automation and very high fixed cost; sugar refining, solvent extraction of oil, flour milling and large dairy pasteurisation lines are close to this type.
As one moves from job to continuous, unit variable cost falls, capital intensity and fixed cost rise, flexibility falls, work-in-process falls, labour skill shifts from craft to monitoring, and planning moves from scheduling individual orders to balancing flows. A distinct classification is by customer order: make-to-stock (flour, sugar), make-to-order (custom feed formulations), and assemble-to-order (gift hampers assembled from stocked items).
Robert Hayes and Steven Wheelwright (1979) linked product life cycle with process life cycle in the product-process matrix. Product structure runs along one axis, from low-volume, one-of-a-kind products to high-volume, standardised commodities. Process structure runs along the other, from jumbled flow (job shop) through disconnected line flow (batch) and connected line flow (assembly line) to continuous flow. Firms positioned on the diagonal have matched their process to their product: a job shop for custom products, continuous flow for commodities.
Off-diagonal positions carry costs. A firm using a flexible job-shop process for a high-volume commodity bears high out-of-pocket costs because it cannot achieve the unit costs of rivals using dedicated lines. A firm using a rigid continuous process for low-volume, varied products bears opportunity costs because it cannot respond to customers who want variety. The matrix also describes evolution: as a new product such as a ready-to-eat millet snack moves from introduction to maturity, its volume rises and the firm should move its process down the diagonal from batch towards line flow.
Product design decides what the customer receives: formulation, pack size, shelf life and specification. Process design decides how it is made. Concurrent (simultaneous) engineering brings marketing, design, production, quality and suppliers into the design team from the start, which shortens development time and avoids products that are hard to make.
Quality function deployment (QFD) translates customer requirements, the voice of the customer, into technical characteristics. Its main tool, the house of quality, is a matrix that relates what customers want (crisp texture, low oil, long shelf life) to how the product is engineered (frying temperature, moisture content, pack atmosphere), with a roof that shows how the technical characteristics interact.
Robust design, associated with Genichi Taguchi, aims for a product whose performance is insensitive to variation in raw material or conditions, which matters when raw produce varies by season. Process design uses flow charts and process maps to set the sequence of operations, choose equipment and fix control points, and in food plants it is also the starting point for hazard analysis.
A location decision commits capital for decades and fixes a large share of transport, labour and raw-material costs. For agri-processing it is often decided by raw material. Alfred Weber's theory of industrial location (1909) classified materials as ubiquitous (available everywhere, such as water in many places) or localised, and as pure (whole weight enters the product) or weight-losing. His material index is the weight of localised materials divided by the weight of the product. When it is greater than one the process is weight-losing and the plant should be near the raw material; when it is less than one, or the product gains weight from ubiquitous inputs, it should be near the market. Sugar mills, cotton ginneries, tea factories and rice mills are weight-losing and perishable, so they sit near the crop; soft-drink bottling, which adds water, tends to sit near consumers.
Location factors are grouped as regional (raw material supply, market, transport links, climate, power, water), community (labour availability and skills, attitudes, taxes, incentives, services) and site (land cost, soil and drainage, room for expansion, effluent disposal, zoning).
The factor-rating method handles qualitative and quantitative factors together. List the relevant factors, give each a weight so that weights sum to one, score each site on each factor on a common scale, multiply and add; the highest total wins. Example: a fruit-pulp plant weighs raw-material proximity 0.40, labour 0.20, market access 0.25 and power 0.15. Site A scores 8, 6, 7 and 5; site B scores 6, 9, 8 and 7. A totals 3.2 + 1.2 + 1.75 + 0.75 = 6.90; B totals 2.4 + 1.8 + 2.0 + 1.05 = 7.25, so B is preferred despite being farther from the orchards.
The centre-of-gravity method finds a location that minimises distribution cost to or from several points, treating cost as proportional to distance and volume. Place the points on a grid; the chosen X coordinate is the sum of each point's X times its volume, divided by total volume, and likewise for Y. Example: collection centres at (2, 8), (6, 3) and (9, 7) send 300, 500 and 200 tonnes. X = (600 + 3,000 + 1,800)/1,000 = 5.4; Y = (2,400 + 1,500 + 1,400)/1,000 = 5.3.
Locational cost-volume (break-even) analysis compares sites whose fixed and variable costs differ. Total cost at a site equals its annual fixed cost plus variable cost per unit times volume. Plot each line; the site with the lowest line at the expected volume is chosen, and the crossover volume between two sites is where their costs are equal: difference in fixed cost divided by difference in variable cost per unit.
Example: a dal mill considers three sites. Site A has fixed cost Rs 30 lakh a year and variable cost Rs 400 per tonne; site B Rs 50 lakh and Rs 300; site C Rs 20 lakh and Rs 480. At 15,000 tonnes a year, total costs are A: Rs 90 lakh, B: Rs 95 lakh, C: Rs 92 lakh, so A is cheapest. The crossover between C and A is 10,00,000/80 = 12,500 tonnes; between A and B it is 20,00,000/100 = 20,000 tonnes. So C is best below 12,500 tonnes, A between 12,500 and 20,000, and B above 20,000.
Layout is the physical arrangement of departments, machines, storage and people. A product (line) layout arranges equipment in the order of operations for one product or a family of similar products; a milk-pouch line or a flour mill is an example. It gives low handling cost, low work-in-process and simple control, but is inflexible and vulnerable, because one stopped machine halts the line. A process (functional) layout groups similar machines together, such as all dryers in one bay and all grinders in another; it suits job and batch work with varied routes, offers flexibility and uses general-purpose equipment, but material travels farther, work-in-process is higher and scheduling is harder. A fixed-position layout keeps the product still and brings workers and equipment to it, as in erecting a large silo or a greenhouse. A cellular layout, based on group technology, forms cells of dissimilar machines that make a family of similar parts or products, combining some of the flow of a line with some flexibility of a process layout.
Tools include flow process charts, the from-to (travel) chart, relationship charts that rate the desirability of closeness between departments, and computer methods. The load-distance method for a process layout multiplies the number of trips (or loads) between each pair of departments by the distance between them and sums; the layout with the smaller total is better.
Line balancing assigns the tasks of a product layout to workstations so that each station's work fits within the cycle time and idle time is minimised. The cycle time is the maximum time a unit may spend at each station, set by demand: available production time per period divided by required output per period. The theoretical minimum number of stations is the sum of task times divided by the cycle time, rounded up. Line efficiency is the sum of task times divided by (number of stations times cycle time); balance delay is one minus efficiency. The cycle time cannot be shorter than the longest single task unless that task is split or duplicated.
Worked example. A mango-pulp canning line must pack 900 cans in a shift with 450 productive minutes (27,000 seconds), so cycle time is 30 seconds. Seven tasks take A 12, B 18, C 10, D 22, E 15, F 8 and G 25 seconds, total 110 seconds. Precedence: A comes first; B and C follow A; D follows B; E follows C; F follows both D and E; G follows F. The theoretical minimum is 110/30 = 3.67, so 4 stations. Using the ranked positional weight method (Helgeson and Birnie), each task's weight is its own time plus the times of all tasks that follow it: A 110, B 73, C 58, D 55, E 48, F 33, G 25. Assigning in weight order while respecting precedence and cycle time gives stations {A, B} 30 s, {C, E} 25 s, {D, F} 30 s and {G} 25 s. Four stations, efficiency 110/120 = 91.7 per cent, balance delay 8.3 per cent. The simpler longest-operation-time rule on the same data yields five stations and 73.3 per cent efficiency, which shows that heuristics differ and none guarantees the optimum.
Every production plan starts with a forecast. Here the concern is the forecast the plant needs for capacity, aggregate and material planning; demand planning across the supply chain is treated in the supply-chain subject.
Qualitative methods are used when data are scarce: executive opinion, sales-force composite, consumer surveys and the Delphi method, in which a panel of experts answers anonymously in rounds with feedback until views converge. Quantitative time-series methods assume the past pattern continues. A simple moving average of n periods averages the last n observations; with demand of 410, 450 and 470 tonnes, the three-period forecast is 443.3 tonnes. Exponential smoothing sets the new forecast equal to the old forecast plus alpha times the last error: with forecast 500, actual 560 and alpha 0.2, the next forecast is 500 + 0.2 x 60 = 512.
Accuracy is measured by mean absolute deviation (MAD, average of absolute errors), mean squared error (MSE) and mean absolute percentage error (MAPE). Bias shows up as errors of one sign. The tracking signal is the running sum of forecast errors divided by MAD; when it drifts outside limits chosen by the firm (often about plus or minus 4 in textbook examples) the model should be reviewed. Example: errors of 20, -20, 10, 60 and 90 give MAD 40 and a tracking signal of 160/40 = 4, a sign that demand is rising faster than the forecast.
Aggregate planning sets overall output, workforce and inventory levels for a medium horizon, typically 3 to 18 months, in aggregate units (tonnes, litres, standard cases) rather than individual products. Capacity (supply) options: hire or lay off, overtime or idle time, part-time or seasonal workers, subcontracting, and building or drawing down inventory. Demand options: pricing, promotion, back-ordering and developing counter-seasonal products; an ice-cream plant that makes frozen snacks in winter is using the last.
Three pure strategies are taught. A level strategy produces at a constant rate and absorbs demand variation with inventory or back orders, which keeps the workforce stable but carries holding cost. A chase strategy changes production each period to equal demand, by hiring and laying off or by varying hours, which keeps inventory low but incurs hiring, training and layoff costs and can harm morale and quality. A mixed (hybrid) strategy combines them, and is usually cheapest in practice.
Example: a potato-chip plant faces quarterly demand of 3,000, 5,000, 7,000 and 5,000 tonnes, total 20,000. A level plan makes 5,000 each quarter; ending inventories are 2,000, 2,000, 0 and 0, so at Rs 20 per tonne per quarter holding cost is Rs 80,000. A chase plan, starting from a previous rate of 5,000, cuts output by 2,000 in quarter 1, raises it by 2,000 in each of quarters 2 and 3, and cuts it by 2,000 in quarter 4. If raising the rate costs Rs 50 per tonne and cutting it Rs 80 per tonne, the chase plan costs 4,000 x 50 + 4,000 x 80 = Rs 5,20,000, so the level plan is far cheaper here. In agri-processing the raw-material side must be planned too: a plant cannot chase demand if the crop is not in season, so level production of a storable intermediate is common.
The master production schedule (MPS) disaggregates the aggregate plan into specific end products, quantities and time periods (usually weeks). Rough-cut capacity planning checks whether the MPS is feasible against critical resources, such as the evaporator in a milk-powder plant or the retort in a canning line, before it is released. Time fences protect stability: inside the frozen zone (the nearest weeks) changes need senior approval because material is already committed; in the slushy zone limited changes are allowed; in the liquid zone further out the schedule is freely changed. Available-to-promise (ATP) tells sales how much of the scheduled output is still uncommitted: in the first period it equals on-hand stock plus the MPS quantity minus customer orders until the next MPS receipt; in later periods it is the MPS quantity minus orders until the next receipt. Example: with 50 cases on hand, an MPS of 200 cases in week 1, the next MPS receipt in week 3, and orders of 120 in week 1 and 80 in week 2, ATP in week 1 is 50 + 200 - 200 = 50 cases. The MPS horizon must at least cover the cumulative lead time of the product.
Scheduling assigns specific jobs to machines and times. Priority rules sequence jobs waiting at one machine. First come, first served (FCFS) is simple and seen as fair. Shortest processing time (SPT) takes the shortest job first; for a single machine it minimises mean flow time, mean waiting time and the average number of jobs in the system. Earliest due date (EDD) takes the job due soonest; it minimises the maximum lateness. The critical ratio (CR) is time remaining until the due date divided by work remaining; a ratio below one means the job is behind schedule and it goes first. The Moore (Hodgson) algorithm minimises the number of late jobs.
Example: a custom flour mill has five orders with processing times and due dates (days): P 6 and 10, Q 2 and 6, R 8 and 18, S 3 and 15, T 9 and 23. FCFS in the order P, Q, R, S, T gives completion times 6, 8, 16, 19, 28, mean flow time 15.4 days and mean tardiness 2.2 days. SPT (Q, S, P, R, T) gives mean flow time 13.0 and mean tardiness 1.4. EDD (Q, P, S, R, T) gives mean flow time 13.6, mean tardiness 1.2 and only two late jobs. All three sequences finish at day 28 and all have a maximum lateness of 5 days here, because the last job T ends late whatever the order. The average number of jobs in the system equals total flow time divided by makespan: 65/28 = 2.32 under SPT. No rule is best on every measure, so the choice follows what matters to customers.
When every job passes through the same two machines in the same order, S. M. Johnson's rule (1954) minimises makespan, the time to finish all jobs. List the time of each job on each machine. Find the smallest time anywhere. If it is on the first machine, schedule that job as early as possible; if on the second, as late as possible. Remove the job and repeat; break ties arbitrarily.
Example: five seed lots must be dried (machine 1) and then graded (machine 2). Times in hours: J1 4 and 6, J2 8 and 3, J3 2 and 7, J4 6 and 5, J5 5 and 9. The smallest time is J3's 2 on the dryer, so J3 goes first. Next is J2's 3 on the grader, so J2 goes last. Then J1's 4 on the dryer goes second. Then J4's 5 on the grader and J5's 5 on the dryer tie; J4 goes to the latest open slot and J5 to the earliest. The sequence J3, J1, J5, J4, J2 finishes in 32 hours, against 36 hours in the order J1 to J5.
Work study is the systematic examination of human work to improve how it is done and to set standards for how long it should take. It has two linked parts: method study, which finds a better way, and work measurement, which times it. The founders are Frederick W. Taylor, who used stopwatch time study and published The Principles of Scientific Management in 1911, and Frank and Lillian Gilbreth, who studied motions and classified basic hand motions as therbligs (their name spelled nearly backwards).
The standard method-study procedure, as set out by the ILO, runs: select the job worth studying; record the present method; examine it critically; develop a better method; evaluate, define and install it; and maintain it.
Flow process charts, for a worker, material or equipment, use five symbols from the American Society of Mechanical Engineers: operation (circle), inspection (square), transport (arrow), delay (a letter D shape) and storage (triangle).
Work measurement establishes the time a qualified worker needs to do a specified job at a defined level of performance. In time study the job is broken into elements, each timed over several cycles. The analyst rates the worker's pace against a standard rating (commonly 100) to convert observed time to basic or normal time: normal time equals observed time times rating divided by 100. Allowances are then added for relaxation (personal needs and fatigue), contingencies and sometimes policy, giving standard time. Example: a worker packs a 5 kg bag in an average observed 2.0 minutes at a rating of 110; normal time is 2.2 minutes. If allowances are 15 per cent of normal time, standard time is 2.2 x 1.15 = 2.53 minutes. Standard output in a 480-minute shift at a standard time of 2.4 minutes is 200 units.
Work sampling (activity sampling), introduced by L. H. C. Tippett, takes many random instantaneous observations to estimate the proportion of time spent on each activity, such as the share of time a harvester or a forklift is idle. The sample needed for an absolute accuracy e at a confidence level with normal value z is n = z squared times p(1 - p) divided by e squared. With p = 0.25, e = 0.05 and z = 1.96, n = 288.1, so 289 observations. If the accuracy is instead specified as plus or minus 5 per cent of p (relative), e becomes 0.0125 and n rises to about 4,610.
Materials management is the integrated control of materials from purchase to the point of use: purchasing, receiving, inspection, storage, issue, inventory control, handling and disposal of scrap and surplus. In most agri-processing plants purchased raw material is the largest single cost, though its share varies by product and plant, so this function drives profit.
Inventory is any stored resource held for future use. By form it is raw material (paddy, raw milk held briefly, oilseed), work-in-process (partly milled rice, curd setting), finished goods (packed flour) and maintenance, repair and operating supplies (spares, lubricants, cleaning chemicals). By function it is cycle stock (from buying in lots), safety or buffer stock (against uncertain demand or supply), anticipation or seasonal stock (built ahead of a peak or bought at harvest when the crop is available and cheap), pipeline or transit stock, and decoupling stock that lets stages work independently. Agri-processors carry unusually large anticipation stocks because raw material arrives in a few months but sales run all year; this ties up working capital and adds storage losses from moisture, pests and quality decline.
Inventory costs are ordering or set-up costs, holding costs (capital, storage, insurance, deterioration, obsolescence) and shortage costs. A plant with cost of goods sold of Rs 12 crore and average inventory of Rs 3 crore turns stock 4 times a year, or about 91 days of inventory.
Not every item deserves equal attention. ABC analysis, based on the Pareto principle, ranks items by annual consumption value (annual usage times unit cost). Typically a small share of items, roughly 10 to 20 per cent, accounts for most of the value, roughly 70 to 80 per cent; these are A items, controlled tightly with frequent review, accurate records and close follow-up. C items, the many items of low value, get simple rules and larger safety stocks. The exact cut-offs are a management choice, not fixed values.
Other classifications cover what ABC misses. VED (vital, essential, desirable) ranks spares by the consequence of a stock-out: a cheap bearing for the main pasteuriser pump can be vital. FSN (fast-moving, slow-moving, non-moving) uses rate of issue and flags obsolete stock. SDE (scarce, difficult, easy) ranks by availability, which matters for imported spares. Combined matrices such as ABC with VED are used in practice: an A-and-vital item gets the closest control. For food ingredients and packaging, first-expired-first-out issue and lot identification on receipt make traceability and recalls possible.
Material requirements planning (MRP), set out in Joseph Orlicky's 1975 book, calculates the quantity and timing of dependent-demand items: components and materials whose demand follows from the schedule of end products. Independent demand, for end products, is forecast; dependent demand is calculated. MRP has three inputs: the MPS, the bill of materials (BOM), which lists the components and quantities in each parent in a product-structure tree, and the inventory status file (on hand, scheduled receipts, lead times, lot sizes).
The logic, for each item from the top level down, is explosion and netting. Gross requirements come from the parent's planned orders times the quantity per parent. Net requirements equal gross requirements minus on-hand stock minus scheduled receipts, plus safety stock if one is kept. Planned order receipts cover net requirements according to the lot-sizing rule (lot-for-lot, fixed quantity or period order quantity), and planned order releases are offset earlier by the item's lead time. Example: gross requirement 500, on hand 120, scheduled receipt 80 and safety stock 50 give a net requirement of 350. For a juice carton of 12 bottles, 200 cartons need 2,400 bottles; with 400 bottles on hand, 2,000 bottles must be made, and if each needs one cap and 300 caps are on hand, 1,700 caps are ordered. MRP suits discrete, assembled and packaged products; it is less natural for continuous processes, where process-flow scheduling is used.
Manufacturing resource planning (MRP II), promoted by Oliver Wight, extends MRP into a closed-loop system for the whole firm. It adds capacity requirements planning (CRP), which converts planned orders into load on each work centre and checks it against capacity, shop-floor control, purchasing, and links to finance and marketing so that the same data drive the production plan, the business plan and the sales plan. The closed loop means execution data feed back to revise the plans. Enterprise resource planning (ERP), a term used from the early 1990s, integrates all business functions on a common database: finance and accounting, human resources, sales and distribution, procurement, production, quality, maintenance and, in modern systems, supply-chain and customer modules.
For an agribusiness, ERP adds value when it handles farm procurement (grower accounts, quality-based payment on fat and solids-not-fat in milk or on sugar content in cane), batch and lot traceability, shelf-life and expiry management, and variable yields, all of which general manufacturing packages may not cover without adaptation.
Just-in-time (JIT) produces what is needed, when it is needed, in the quantity needed. It grew out of the Toyota Production System developed by Taiichi Ohno and colleagues from the 1950s. JIT treats inventory as a cost that hides problems: a common image is a river whose water level (inventory) covers rocks (defects, breakdowns, long set-ups); lowering the level exposes the rocks so they can be removed.
Elements of JIT include pull production, in which each stage makes only what the next stage has used; small lot sizes made possible by quick set-ups; level (mixed-model) scheduling, called heijunka; takt time, the available time divided by customer demand, which sets the pace of production; jidoka, stopping the line when an abnormality occurs; close, long-term relations with few suppliers who deliver small, frequent lots of assured quality; and continuous improvement (kaizen). Shigeo Shingo's single-minute exchange of die (SMED) method cuts changeover time by separating internal set-up (done only while the machine is stopped) from external set-up (done while it runs) and converting internal to external work. His poka-yoke (mistake-proofing) devices make errors impossible or immediately visible.
JIT has limits in agribusiness. Raw produce arrives at harvest, not on demand, and cannot always be pulled in small lots; long-distance or monsoon-disrupted supply calls for buffers; and perishables cannot wait in queues. JIT principles therefore apply most readily downstream of the raw-material store: in packing, ingredient supply, packaging materials and distribution.
Kanban (Japanese for card or signboard) is the signal system that makes pull work. A withdrawal (conveyance) kanban authorises moving a container of parts from the supplying stage to the using stage; a production kanban authorises the supplying stage to make a container to replace one withdrawn. Nothing is produced or moved without a kanban, so the number of cards caps work-in-process.
The number of kanbans is N = D x L x (1 + alpha) / C, where D is demand per unit time, L is the lead time (waiting plus processing plus conveyance) in the same time unit, alpha is a safety factor and C is container capacity. Example: a pouch-filling line uses 200 packs of spice premix per hour; replenishment lead time is 1.5 hours; safety factor 0.2; each tote holds 40 packs. N = 200 x 1.5 x 1.2 / 40 = 9 kanbans. Managers then remove kanbans deliberately, one at a time, to expose problems.
Lean is the broader philosophy drawn from the Toyota system and named by researchers at the Massachusetts Institute of Technology in the late 1980s; James Womack, Daniel Jones and Daniel Roos popularised it in The Machine That Changed the World (1990). Womack and Jones's Lean Thinking (1996) gave five principles: specify value from the customer's view; identify the value stream and remove steps that do not create value; make value flow; let the customer pull; and pursue perfection.
Lean distinguishes muda (waste), mura (unevenness) and muri (overburden). The seven wastes identified by Ohno are overproduction, waiting, transport, over-processing, inventory, unnecessary motion and defects; many writers add an eighth, unused human talent.
Lean tools include value stream mapping, which draws material and information flow with times and highlights the small share of lead time that adds value; 5S workplace organisation (sort, set in order, shine, standardise, sustain); visual management; standard work; kaizen events; and total productive maintenance.
Business process re-engineering (BPR) was defined by Michael Hammer and James Champy (1993) as the fundamental rethinking and radical redesign of business processes to achieve dramatic improvements in critical, contemporary measures of performance such as cost, quality, service and speed. Hammer's earlier article (1990) urged firms not to automate existing processes but to obliterate them. BPR differs from kaizen: kaizen is incremental, continuous and bottom-up; BPR is radical, one-time and top-down.
Value engineering (VE), or value analysis, was developed by Lawrence D. Miles at General Electric in the late 1940s. Value is the ratio of function to cost; VE seeks to provide the required function at the lowest total cost without reducing quality, reliability or safety. The job plan runs through information, function analysis, creative (brainstorming alternatives), evaluation, development and presentation or implementation phases. Functions are described in two words, a verb and a noun ("protect product", "contain liquid"), and classified as basic or secondary. A snack maker might find that a thinner laminate still gives the barrier needed to protect the product, reducing packaging cost without affecting shelf life; that finding must be confirmed by shelf-life testing, not assumed.
Quality has several accepted definitions. Joseph Juran called it fitness for use; Philip Crosby, conformance to requirements; the ISO 9000 vocabulary, the degree to which a set of inherent characteristics of an object fulfils requirements. Walter Shewhart developed the control chart at Bell Telephone Laboratories in the 1920s and the plan-do-check-act cycle, which W. Edwards Deming taught (he preferred plan-do-study-act). Deming's 14 points include creating constancy of purpose, ceasing dependence on mass inspection by building quality in, ending the practice of awarding business on price alone, improving constantly, instituting training, driving out fear, breaking down barriers between departments and removing barriers to pride of workmanship. Juran's trilogy is quality planning, quality control and quality improvement. Crosby promoted zero defects and argued that quality is free because the cost of poor quality exceeds the cost of prevention. Kaoru Ishikawa promoted quality circles and the cause-and-effect (fishbone) diagram. Genichi Taguchi proposed the quality loss function, under which any deviation from target, even within specification, imposes a loss that grows roughly with the square of the deviation.
Total quality management (TQM) is an organisation-wide approach built on customer focus, continuous improvement, employee involvement, process management, fact-based decisions and supplier partnership.
The cost of quality has four categories. Prevention costs are spent to stop defects: quality planning, training, supplier development, process design, preventive maintenance. Appraisal costs are spent to find defects: inspection, testing, laboratory analysis, calibration. Internal failure costs arise from defects found before delivery: scrap, rework, downgrading of product, re-inspection, lost yield. External failure costs arise after delivery: returns, complaints, recalls, warranty, lost customers. The TQM argument is that spending more on prevention reduces the much larger failure costs. Example: if a dairy spends Rs 2 lakh on prevention, Rs 3 lakh on appraisal, Rs 6 lakh on internal failure and Rs 9 lakh on external failure, failure costs are 15 of the 20 lakh total, or 75 per cent, a sign that prevention is underfunded.
Ishikawa's seven basic quality tools are the check sheet, histogram, Pareto chart (to separate the vital few causes from the trivial many), cause-and-effect diagram (with bones often grouped as man, machine, method, material, measurement and environment), scatter diagram, control chart and either stratification or a flow chart, depending on the author.
Six Sigma, developed in an American electronics firm in the 1980s, aims at processes so capable that defects fall to 3.4 per million opportunities; that figure assumes the process mean may drift by 1.5 standard deviations over time. Projects follow DMAIC: define, measure, analyse, improve, control. Defects per million opportunities (DPMO) equals defects divided by (units times opportunities per unit), times one million: 18 defects in 500 packs with 6 opportunities each gives 18/3,000 x 1,000,000 = 6,000 DPMO.
Statistical process control (SPC) uses sampling and control charts to tell whether a process is stable. All processes vary. Common (chance, random) causes are the many small sources built into the process; special (assignable) causes are identifiable events such as a worn blade, a new raw-material lot or an untrained operator. A process with only common-cause variation is in statistical control. Control limits describe what the process does, usually at the mean plus and minus three standard deviations of the plotted statistic; specification limits describe what the customer requires. The two are different and should never be confused.
Variables charts are for measured characteristics. The X-bar chart tracks subgroup means and the R chart tracks subgroup ranges. With grand mean X-double-bar and average range R-bar, limits are X-double-bar plus or minus A2 times R-bar for the X-bar chart, and D3 x R-bar and D4 x R-bar for the R chart. For subgroups of five, standard tables give A2 = 0.577, D3 = 0 and D4 = 2.114. Example: a 500 g flour-pack filler with R-bar 8 g gives X-bar limits of 504.62 g and 495.38 g, and an R chart upper limit of 16.91 g with lower limit zero. The process standard deviation can be estimated as R-bar/d2, with d2 = 2.326 for subgroups of five.
Attributes charts are for counts. The p chart plots the proportion defective with limits p-bar plus or minus 3 times the square root of p-bar(1 - p-bar)/n; a negative lower limit is set to zero. The np chart plots the number defective; the c chart plots defects per unit of constant size with limits c-bar plus or minus 3 times the square root of c-bar; the u chart handles varying unit size. A point outside the limits signals a likely special cause. With three-sigma limits, about 0.27 per cent of points from a stable normal process fall outside by chance.
Process capability compares natural spread with specification width. Cp = (USL - LSL)/(6 sigma); Cpk = the smaller of (USL - mean) and (mean - LSL), divided by 3 sigma. A centred process has Cp = Cpk; Cpk lower than Cp means the process is off-centre. Many buyers ask for Cpk of at least 1.33, but the requirement is set by the customer. Capability is meaningful only for a process already in control.
Acceptance sampling decides whether to accept or reject a lot on the basis of a sample. It judges lots; it does not control the process, which is SPC's job. A single sampling plan by attributes is defined by sample size n and acceptance number c: take n items; accept the lot if defectives are c or fewer.
The operating characteristic (OC) curve plots the probability of accepting a lot against its fraction defective. The acceptable quality level (AQL) is the worst quality the consumer considers acceptable as a process average; the lot tolerance percent defective (LTPD), or rejectable quality level, is the poor quality the consumer wants rejected. Producer's risk (alpha) is the chance of rejecting a good lot at the AQL; consumer's risk (beta) is the chance of accepting a bad lot at the LTPD. Values of 5 per cent and 10 per cent are common textbook choices, not rules. Example: for n = 50 and c = 1 with lots 2 per cent defective, the binomial probability of acceptance is 0.98^50 + 50 x 0.02 x 0.98^49 = 0.736; the Poisson approximation with np = 1 gives 2e^-1 = 0.736.
With rectifying inspection, rejected lots are screened and defectives replaced, so the average outgoing quality (AOQ) can never exceed a maximum, the average outgoing quality limit (AOQL).
A standard is a documented agreement on specifications or procedures, set by a recognised body, that ensures products and processes are fit for purpose. In India, the Bureau of Indian Standards (BIS) is the national standards body under the Bureau of Indian Standards Act, 2016, which replaced the 1986 Act; it succeeded the Indian Standards Institution founded in 1947. BIS formulates Indian Standards and runs product certification, whose mark is the ISI mark. Certification is voluntary for most products but compulsory for products notified by the government under quality control orders; the notified list changes, so it must be checked.
AGMARK is the certification mark for agricultural and allied produce under the Agricultural Produce (Grading and Marking) Act, 1937, administered by the Directorate of Marketing and Inspection under the Ministry of Agriculture. It certifies grade and quality against grade standards notified for commodities such as pulses, cereals, spices, edible oils, ghee, honey and others; for most commodities it is voluntary, though particular rules have at times made it compulsory. The Food Safety and Standards Authority of India (FSSAI), under the Food Safety and Standards Act, 2006, sets mandatory food-safety standards and licenses food businesses.
Other countries have their own equivalents: national grade standards for grains, meat and produce, marketing standards in the European Union, and at the international level the Codex Alimentarius Commission of FAO and WHO, whose food standards are the reference point for international trade disputes under the WTO Agreement on the Application of Sanitary and Phytosanitary Measures.
The International Organization for Standardization (ISO), based in Geneva, publishes international standards through national member bodies. Management-system standards specify what an organisation's system must do, not how its product must be made, and they are certified by accredited third-party certification bodies.
ISO 9001 specifies requirements for a quality management system (QMS). The current edition is ISO 9001:2015, supported by ISO 9000 (fundamentals and vocabulary). It rests on seven quality management principles: customer focus, leadership, engagement of people, process approach, improvement, evidence-based decision making and relationship management. ISO 14001 covers environmental management systems, ISO 45001 occupational health and safety, and ISO 22000 food safety management systems. ISO 22000 (current edition 2018) combines interactive communication along the food chain, system management, prerequisite programmes and HACCP principles, and applies to any organisation in the food chain, from farm to retail. Certification shows that a system exists and is followed; it does not guarantee that every product is safe or perfect.
Hazard Analysis and Critical Control Point (HACCP) is a preventive system that identifies, evaluates and controls hazards significant for food safety. It was developed in the 1960s to supply safe food for the United States space programme and is now set out by the Codex Alimentarius in its General Principles of Food Hygiene (CXC 1-1969, revised). Hazards are biological, chemical (including residues, mycotoxins and allergens) or physical (glass, metal, stones). HACCP rests on prerequisite programmes, the good hygiene and manufacturing practices (cleaning, pest control, water quality, personal hygiene, maintenance) that must be in place first.
The seven principles are: conduct a hazard analysis and identify control measures; determine the critical control points (CCPs); establish validated critical limits; establish a system to monitor control of CCPs; establish corrective actions when monitoring shows a deviation; validate the plan and establish verification procedures to confirm it works; and establish documentation and record keeping. Codex places these principles within a 12-step logic sequence: assemble the HACCP team, describe the product, identify intended use and users, construct a flow diagram, confirm it on site, and then the seven principles as steps 6 to 12.
A CCP is a step at which control is essential to prevent or eliminate a hazard or reduce it to an acceptable level, and after which no later step will do so. In milk processing, pasteurisation is the classic CCP for vegetative pathogens; its critical limits are a time-temperature combination set by national regulation, and a flow-diversion valve returns under-heated milk automatically as the corrective action. Metal detection before packing is a common CCP for physical hazards.
Maintenance keeps equipment able to perform its function. In a seasonal plant an in-season breakdown can cost far more than the repair, because raw material spoils, so major work is done in the off-season.
Breakdown (corrective, reactive) maintenance repairs equipment only after it fails. It suits cheap, non-critical items with spare capacity. Preventive (planned, scheduled) maintenance services equipment at fixed intervals of time or usage, such as lubricating, replacing belts and filters, and overhauling pumps before the season. Predictive or condition-based maintenance monitors the actual condition of equipment through vibration analysis, thermography, oil analysis or ultrasound, and intervenes when a measured signal shows deterioration; it avoids both unexpected failure and unnecessary replacement.
The choice is economic. As preventive effort rises, its cost rises and breakdown cost falls; total cost is lowest at an intermediate level. Example: 20 machines under a breakdown policy fail on average 0.5 times each per month at Rs 6,000 per breakdown, costing Rs 60,000 a month. A preventive programme costing Rs 1,000 per machine per month cuts failures to 0.1 per machine per month; its total cost is Rs 20,000 + 20 x 0.1 x 6,000 = Rs 32,000, a saving of Rs 28,000 a month.
Failure behaviour over equipment life is often described by the bathtub curve: a falling failure rate in the early (infant mortality or burn-in) period, a roughly constant rate in the useful-life period, and a rising rate in the wear-out period.
Mean time between failures (MTBF) is operating time divided by the number of failures; mean time to repair (MTTR) is total repair time divided by the number of repairs. Inherent availability is MTBF/(MTBF + MTTR): with MTBF 190 hours and MTTR 10 hours, availability is 95 per cent. Reliability is the probability that equipment performs its function for a stated period under stated conditions. For components in series, where any failure stops the system, system reliability is the product of component reliabilities: 0.95 x 0.98 x 0.90 = 0.838. For components in parallel (redundancy), the system fails only if all fail: two pumps each of reliability 0.9 give 1 - 0.1 x 0.1 = 0.99.
Total productive maintenance (TPM) was developed in Japan and systematised by Seiichi Nakajima of the Japan Institute of Plant Maintenance. It involves every employee in maintaining equipment, with operators taking on routine cleaning, inspection and lubrication (autonomous maintenance). It is usually described with eight pillars: autonomous maintenance, planned maintenance, focused improvement, quality maintenance, early equipment management, training and education, safety, health and environment, and TPM in administration. TPM targets six big losses: breakdowns, set-up and adjustment, idling and minor stops, reduced speed, process defects, and reduced yield at start-up.
Overall equipment effectiveness (OEE) combines them: OEE = availability x performance x quality. Availability is operating time over planned production time; performance is (ideal cycle time x total count) over operating time; quality is good count over total count. Example: in a 480-minute planned shift with 48 minutes of stops, a packing machine with ideal cycle time 0.5 minutes makes 800 packs, of which 776 are good. Availability is 432/480 = 0.90, performance is 400/432 = 0.926, quality is 776/800 = 0.97, and OEE is 0.808, which equals good count times ideal cycle time over planned time (388/480).
A student should be able to classify any agri-processing unit by process type and layout, and compute the standard numericals quickly and without unit slips: utilisation and efficiency, crossover volume between sites, cycle time and line efficiency, mean flow time under SPT and EDD, a Johnson sequence, standard time, control limits for X-bar, R, p and c charts, Cp and Cpk, availability and OEE. Learn the thinkers and what each proposed, because attribution questions are cheap marks. In internships, draw a flow process chart of a real plant and note where material waits; it teaches more than any table.
A plant or production manager five years in uses this material weekly. Typical tasks: setting the season's aggregate plan against expected procurement, fixing the weekly MPS and promising dates to buyers from ATP, balancing a packing line when a new pack size is launched, running daily X-bar and R charts on fill weight and p charts on seal defects, maintaining the HACCP plan and records for audits, classifying spares by VED so the season is not lost for a cheap part, and tracking OEE and yield by shift. The decisions are about which loss to attack first, and the Pareto chart answers that.
A senior operations head with 25 years makes the structural decisions: where to locate a new plant given crop shifts and policy incentives, whether to add capacity ahead of demand or lag it, how to extend the processing season with new crops or intermediate stocks, which certifications the market will pay for, whether an ERP rollout is ready, and how to redesign procurement from thousands of farmers. The role is to align operations with strategy, choose the competitive priority the plant must win on, and build a culture where operators own quality and maintenance rather than depending on inspection.
A researcher with long field experience asks why well-known operations methods succeed or fail in agri-food settings: how raw-material variability and seasonality change the value of lean and JIT; how capacity and location decisions respond to climate-driven crop shifts; how certification affects smallholder inclusion and export access; how OEE and productivity should be measured when yields vary; and how digital systems change procurement and traceability. Methods include plant-level panel data, efficiency analysis, simulation, optimisation and case studies. Good work states clearly which results come from which setting and resists generalising a single-country finding.
This manual does not yet cover: the full tables of control-chart constants beyond subgroups of three to five; operating-characteristic curve construction for double and sequential plans; simulation and optimisation methods for job-shop scheduling; detailed product-specific AGMARK grade designations, BIS quality control orders and FSSAI licensing categories, which change and must be checked on the regulators' current lists; country-by-country food-safety law beyond naming the regulators; process engineering of individual agri-processing operations (drying, milling, pasteurisation), which belongs to food and agricultural engineering; and service operations. EOQ, reorder points, queuing and linear programming are in the operations-research subject; transport, warehousing, network design and demand planning are in the supply-chain subject.
Original practice questions written to the syllabus and pattern of the exams named in each question's examTags. They are not past papers and do not reproduce any board's questions. Every numerical answer was computed by script and the working is shown in the explanation.
These notes are free to read here, with no account.
What the Laisha app adds:
Free for ten days, then a subscription. We would rather write that here than have you find it later.