385. Part-Period Balancing

A lot-sizing heuristic (DeMatteis, 1968) based on the EOQ balance condition: at the EOQ optimum, ordering cost equals holding cost. Replicate this for time-varying demand.

385.1. EOQ balance recap

For stationary demand:

EOQ optimality: these are equal. The Part-Period Balancing analog: equate setup cost to part-periods (cumulative holding cost).

385.2. Algorithm

Economic part-period (EPP):

EPP=𝐾

— the number of unit-periods that would generate holding cost equal to one setup. (One unit held for one period contributes of holding cost; total holding cost on EPP unit-periods is EPP=𝐾.)

For an order starting in period 𝑡 covering 𝑘 periods:

Part-periods(𝑘)=𝑗=1𝑘1𝑗𝑑𝑡+𝑗

Extend 𝑘 until part-periods approaches but doesn’t exceed EPP. Stop at the 𝑘 where:

Part-periods(𝑘+1)>EPPbutPart-periods(𝑘)EPP

385.3. Worked example

𝑑=(60,100,80,50,40,70), 𝐾=100, =1. EPP=1001=100.

Start period 1:

Order covers 2 periods.

Start period 3:

Take 𝑘=2 (130 vs 50 — closer to 100? 130 vs 50: |130100|=30,|50100|=50. Take 𝑘=3.). Note: rules differ; some variants pick “closest” rather than “last not exceeding”.

Start period 6: only one period left. Order covers 1 period.

(Worked answer depends on tie-breaking rule.)

385.4. Variants

385.5. Quality

Generally within 5–10% of Wagner-Whitin optimum. Slightly worse than Silver-Meal or LUC in benchmarks, but conceptually simpler — directly mimics EOQ’s balance principle.

385.6. Why it sometimes wins

When demand patterns are close to stationary, PPB hugs the EOQ-like balance more tightly than the period-based or unit-based alternatives.

385.7. See also