Sizing Capacity Before You Know the Peak
Capital, Lead Time and a Two-Year Product Life Cycle
An electronics manufacturer had to commit capital to overseas component capacity before anyone knew how big the product would be. Lead times ran three to six months, and the product would be gone in two years. A model of the supply chain put a cost on being wrong in either direction before the money was spent, and tested what flexible ordering and expediting could recover.
- Sector
- Consumer electronics
- The decision
- How much component capacity to commit to up front
- Scope
- Overseas component plant, ocean freight, US assembly, distribution
- What it was worth
- The cost of over- and under-forecasting, priced before the commitment
- Evidence
- Project record
Three problems that make each other worse
Key components were made overseas and shipped to a US assembly plant, which fed the distribution centers. Loading, ocean transit and unloading, including customs, took 30 days. Three problems fed into each other:
- Demand was hard to forecast. Both the total and the peak of a two-year consumer electronics life cycle are uncertain, and the peak is what sizes the plant.
- Capacity had to be committed in advance. The life cycle is too short to add capacity once the market shows how strong demand really is.
- Lead times were long. At three to six months, the only protection against short-term surprises is a large safety stock at the assembly plant, and that inventory costs money every month it sits there.
Crossing capacity with demand
The model’s inputs were combinations of life-cycle demand and forecast. Three capacity commitments were crossed with three ways actual sales could turn out, including the one where the forecast was right.
Every combination was simulated through the life cycle and scored on capital equipment cost, the cost of prebuilding inventory ahead of a peak the plant could not meet, total shipping cost, inventory at both the overseas plant and the US assembly plant, and customer service.
What each combination did
| Sales A (high) | Sales B (medium) | Sales C (low) | |
|---|---|---|---|
| Capacity 1 (high) | Moderate prebuild during peak | No prebuild; capital somewhat excessive | No prebuild; capital very excessive |
| Capacity 2 (medium) | Massive prebuild during peak | Moderate prebuild during peak | No prebuild; capital somewhat high |
| Capacity 3 (low) | Capacity cannot keep pace with sales | Massive prebuild during peak | Moderate prebuild during peak |
Idle capital Moderate prebuild Massive prebuild Lost sales
Even where capacity matched demand, the model still called for a moderate prebuild during the peak. Committing one level too low turned that into a massive prebuild, and two levels too low meant capacity could not keep pace at all. Committing too high avoided prebuilds and paid for it in idle capital. Prebuilding was the only way to cover a peak the plant could not meet, so a shortfall in capacity turned into inventory built and paid for ahead of the peak.
What flexible ordering and expediting were worth
With capacity fixed, the model tested two ways to live with the forecast error:
Flexible ordering
If the assembly plant could adjust its orders closer to the time of need, it could hold far less safety stock against long lead times. The model asked how much flexibility was enough: flexibility helps the customer, but it pushes cost and risk back onto the supplier. Flexible ordering became the main tool for cutting the large safety stock at the US assembly plant.
Expedited shipments
Air freight answers a demand spike far faster than a month at sea. The model weighed that against its extra shipping cost, alongside the inventory and customer service each option produced.
The same shape of decision turns up well beyond electronics: a seasonal product, a limited-time flavor, a launch with a short window. Wherever capacity has to be committed before demand is known and the peak sizes the plant, the useful question is not “what is the forecast?” but “what does it cost if the forecast is wrong, in each direction?”
Where this sits in ReliaSim
The question runs through the network: a supplier with committed capacity, a long lane, an expedite option, an assembly plant and its safety stock. ReliaSim’s supply chain modeling represents sites, lanes, demand, inventory and ordering in one model, and the simplest version of that chain, a supplier feeding a plant feeding a customer, runs in the browser today.
On attribution. This model was built by Simulation Dynamics, the team behind ReliaSim, on a commercial simulation platform. The account rests on our project record rather than a published paper, and the client is not named. The software in the sandbox did not produce these results.