Logistics systems engineering and supply chain design
This domain tests whether you can model a logistics network as an engineered system - nodes, links, inventory buffers and information flows - and defend design decisions with arithmetic rather than adjectives. Expect questions on postponement, the bullwhip effect, inventory policy, network configuration and total landed cost.
Decoupling point and the bullwhip
- The push–pull boundary (decoupling point) is where demand stops being forecast-driven and becomes order-driven. Moving it upstream buys scale economies; moving it downstream buys responsiveness, which is why postponement (late configuration, merge-in-transit) suits high-variety, short-life products.
- Bullwhip amplification stems from demand-signal processing, order batching, price fluctuation and shortage gaming. Remedies: point-of-sale data sharing, vendor-managed inventory, smaller replenishment batches, allocation on past sales rather than current orders.
- SCOR frames the process as Plan, Source, Make, Deliver, Return and Enable - useful for scoping an answer.
Sizing the buffer
EOQ = sqrt(2DS/H), where D is annual demand, S order cost and H annual holding cost per unit. Reorder point = demand during lead time + safety stock. When both demand and lead time vary:
SS = z · sqrt(L·σd² + d²·σL²)
z comes from the target cycle service level (CSL), which is not the fill rate: CSL is the probability of no stock-out per cycle, fill rate is the fraction of units served from stock.
Worked example
A Kwai Chung spare-parts warehouse has daily demand d = 400 units, σd = 60; lead time L = 9 days, σL = 2 days; target CSL 95% (z = 1.65).
SS = 1.65 × sqrt(9 × 3,600 + 160,000 × 4) = 1.65 × sqrt(32,400 + 640,000) = 1.65 × 820 ≈ 1,353 units.
If lead time were perfectly reliable (σL = 0), SS = 1.65 × 180 = 297 units. Lead-time variability, not demand variability, dominates here - a common exam insight.
Network choices
Hub-and-spoke concentrates flow to raise vehicle fill; direct delivery trades consolidation for speed; cross-docking removes storage but demands schedule discipline; milk runs trade frequency for distance. Decide on total landed cost - freight, duty, insurance, handling, carrying cost, risk - never on freight rate alone.
Traps that cost marks
- Using average lead time inside the safety-stock formula instead of separating mean and variability.
- Confusing CSL with fill rate, or quoting z for 95% as 1.96 rather than 1.65.
- Ignoring inventory carrying cost when comparing sea and air options.
- Presenting an EOQ without checking shelf life, supplier minimum batches or container fill.