How to use it
- 1. Set demandEach store sees Normal daily demand; σ is where multi-echelon gets interesting.
- 2. Place the stockTwo sliders: base stock per store and at the DC. The network diagram shows the split.
- 3. Read serviceStore fill rate is what customers feel; DC fill rate is how often stores get resupplied on time.
- 4. Sweep the curveThe chart shows fill rate across store base-stock levels at your current DC setting.
The network
Demand (per store, daily)
Base-stock levels
Store lead time from DC: 2 days (when the DC has stock - otherwise stores wait). DC lead time from supplier: 5 days, always filled.
Totals
Experiment 1 asks you to hold this total fixed while shifting units between echelons - trade the sliders against each other.
Bubble size shows average on-hand at each node; red halo = the share of days that node stocked out.
Store fill rate vs store base stock, at your current DC level. The marked point is your setting; note the flattening - the last units buy almost nothing.
Same total stock, five DC/store splits: service peaks at a balanced split - all-forward loses pooling, all-at-DC starves the shelves.
Try these experiments
- Keep the total at 400 and shift units between the sliders - find the split that maximises store fill. Neither extreme wins.
- Raise σ to 12: the best split moves toward the DC - pooling is worth more when demand is noisier.
- Starve the DC (S=0) with generous stores: fill holds for a while, then collapses as pipelines empty - DC stockouts hit stores with a delay.
About this tool
The concept
In a multi-echelon chain, each tier's service depends on the tier above. Stock at the DC is shared insurance; stock in stores is committed to one market.
How it works
Every node runs a base-stock policy: order yesterday's demand to restore the level. Store orders wait when the DC is out - that delay is how upstream failure propagates.
Winning strategy
Push enough forward to cover lead-time demand, pool the rest upstream. As variability rises, shift the balance toward the pool.