Question 5 of 5: Materials Handling, Inventory Control and Forecasting

Question 5: Materials Handling, Inventory Control and Forecasting (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) — objectives of materials handling

Materials handling adds cost and risk but no form utility: the part is no more valuable after it has been moved than before. Everything that follows is a consequence of that single fact, and the objectives of a handling system are therefore stated as reductions rather than as achievements.

The first objective is to eliminate handling wherever it can be eliminated — the cheapest move is the one that never happens, which is why handling analysis and layout analysis are done together and why a well-laid-out cell can delete more handling than any conveyor can economise. Where movement is unavoidable, the aim is to minimise the distance travelled and the number of separate moves, and to move material in the largest practical unit load, since the labour and equipment cost of a trip is largely independent of what is carried; palletising, containers and tote bins all exist for this reason. A related aim is to keep material flowing in as straight and as continuous a path as possible, avoiding backtracking and crossing flows, so that handling equipment is utilised rather than waiting.

Beyond distance and load size, the system aims to reduce work in process and throughput time, since material sitting on a conveyor or in a staging area is inventory by another name; to use the building's cubic volume, not merely its floor area, through stacking, racking and overhead conveyors; to protect the material from damage, contamination and loss; and above all to improve safety, because manual handling is the single largest source of lost-time injury in most plants and the substitution of mechanical handling for lifting and carrying is a control at the top of the hierarchy of controls. Two further objectives are managerial: standardise equipment, containers and interfaces so that the fleet is flexible and spares are few, and design the system so that it generates information — bar-coded or RFID-tagged unit loads turn every move into a transaction that updates the inventory record automatically, which is the link between handling and the control system discussed next. Finally the system must be economically justified over its life, counting energy, maintenance and flexibility to future product change, not merely first cost.

Part (b) — what an inventory control system must accomplish, and the vital areas in designing one

An effective inventory control system exists to deliver an agreed level of service at the lowest total cost, and everything it does is in support of that one trade-off. Concretely, it must answer three questions for every item — what to stock, how much to order, and when to order — and it must answer them automatically enough that management attention is spent only on exceptions. It must maintain record accuracy, since a policy computed on a wrong balance is worse than no policy; it must protect against uncertainty in both demand and lead time by holding a deliberate, costed safety stock rather than an accidental one; it must decouple successive stages of supply and production so that a stoppage in one does not immediately propagate; it must report — turns, fill rate, stockout frequency, ageing and obsolescence — so that performance can be judged; and it must surface exceptions: slow movers, items whose usage has changed, and stock that should be written off.

The vital areas to consider when developing a comprehensive system are these. First, demand forecasting and classification: an ABC analysis, because the small fraction of items carrying most of the value deserves tight control and frequent review, while C items should be governed by simple two-bin rules. Second, the choice of control system type — fixed order quantity against fixed order interval, perpetual against periodic review, independent-demand reorder points against dependent-demand MRP, since applying reorder points to a component whose demand is derived from a schedule is a classic and expensive error. Third, the cost parameters: ordering or set-up cost, carrying rate, and the penalty for a stockout, each of which must be measured rather than assumed. Fourth, lead times, their variability and their management, because safety stock is driven as much by lead-time variance as by demand variance. Fifth, service-level policy, set by management as a business decision, not chosen by the analyst. Sixth, record integrity and physical control: cycle counting, secure and sensibly located stores, unit-load identification and transaction discipline. Seventh, integration with purchasing, production planning, accounting and the handling system, so that one transaction updates every record. Eighth, organisational responsibility: a named owner for inventory investment, with authority over both the policy and the exceptions. And ninth, review of obsolescence and disposal, together with the item-master hygiene that keeps dead items from being reordered.

Part (c) — factors influencing the selection of a forecasting model

No forecasting technique is best in general; the choice is a matching exercise between the situation and the method. The factors that decide it are, first, the time horizon: short-term operational forecasts of days or weeks favour simple time-series methods such as exponential smoothing, while long-term capacity and facility decisions call for causal regression, econometric models or qualitative approaches such as Delphi. Second, the pattern in the data — level, trend, seasonality, cycle or randomness — because a method must contain a term for every component present: simple smoothing on a trended series lags permanently, and Winters' method is wasted on a series with no seasonality. Third, the availability, quantity and quality of historical data; a new product has none, which is why analogy, market survey and executive judgement dominate at launch, and a seasonal model needs several complete cycles of history before it can be fitted.

Fourth, the accuracy required and the cost of being wrong: an item whose stockout shuts a line justifies effort that a stationery item does not. Fifth, the cost of the forecast itself — data collection, software, and analyst time — weighed against that benefit, which for the thousands of items in a stock file usually forces a simple, automatic method. Sixth, the number of items and the time available: a method applied to twenty thousand SKUs overnight must be self-adaptive and cheap, which is the argument for adaptive-response-rate smoothing and for focus forecasting. Seventh, the stability of the environment: where structural change, promotions, competitor action or regulation dominate, extrapolating history is unsound whatever the fit, and a causal or judgemental approach is needed. Eighth, the availability of leading indicators or causal variables, which is what makes regression or an econometric model possible at all. And ninth, the simplicity and transparency of the method to the people who must use it: a forecast that management does not understand will be overridden, so an intelligible method that is used beats a sophisticated one that is not.