Rough-Cut Overload on the CNC Cell: Which Fix, and What the Coarse Model Missed
A machining cell with two CNC centres is the constraint for two MPS families, A and B. Each machine runs two 8-hour shifts, five days a week, giving 160 hours per week before a 10% planned maintenance allowance, so effective capacity is 144 hours per week. Standard hours per unit are 1.5 for family A and 2.2 for family B. The four-week MPS is: W1 A 60 / B 20; W2 A 70 / B 30; W3 A 50 / B 35; W4 A 40 / B 30. A 16-hour planned maintenance shutdown in W3 is not reflected in the 144-hour figure. Overtime is capped at 10% of base hours. The cell also machines shaft SH-12 for family A, and MRP shows planned orders for 60 shafts in W2 at 0.4 hours each - the shafts are not included in the rough-cut load. Sales has just won a request for 20 more B in W2.
1. Build the rough-cut load profile and identify the problem periods, including the items the coarse model omitted.
Read the worked answer
Rough-cut load (family standard hours only):
- W1: 60 x 1.5 = 90; 20 x 2.2 = 44; total 134 h of 144 = 93%
- W2: 70 x 1.5 = 105; 30 x 2.2 = 66; total 171 h = 119%, overload +27 h
- W3: 50 x 1.5 = 75; 35 x 2.2 = 77; total 152 h against effective capacity of 128 h after the 16-hour shutdown = 119%, overload +24 h
- W4: 40 x 1.5 = 60; 30 x 2.2 = 66; total 126 h = 88%, slack 18 h
Corrections the coarse model missed:
- The W3 shutdown cuts capacity from 144 to 128, turning a mild overrun into a 24-hour overload.
- Shafts are real load on the same cell: 60 x 0.4 = 24 h in W2, taking W2 to 195 h, 135% of capacity, overload +51 h.
- The new sales request adds 20 x 2.2 = 44 h to W2 if accepted, pushing W2 to 239 h, 166%.
Aggregate check: family load is 134 + 171 + 152 + 126 = 583 h against 560 h of capacity after the shutdown (576 before). The horizon is short by 23 h in total, and short by far more in the specific weeks where it matters. This is a load-distribution and component-inclusion problem, not just an arithmetic one.
2. Recommend a feasible plan with explicit quantities and hours, and state which modelling assumptions you are relying on.
Read the worked answer
Total overload (with shafts, without the sales request) is 51 h in W2 and 24 h in W3 = 75 h, against available slack of 10 h in W1 and 18 h in W4 = 28 h. Shifting alone cannot solve it; roughly 47 hours must come from overtime, subcontract or demand change. A workable plan:
- Subcontract the 60 shafts (approved vendor, +18% cost, 15-day lead time): frees 24 h in W2, reducing it to 171 h.
- Move 5 units of B from W2 to W4 (5 x 2.2 = 11 h): W2 becomes 160 h; W4 becomes 137 h, within 144.
- Apply 16 h overtime in W2 (the full 10% cap): W2 lands at 144 h - exactly at capacity, no cushion.
- W3 at 152 h vs 128 h: move 4 units of B to W1 (8.8 h; W1 becomes 142.8 h) and 3 units of B to W4 (6.6 h; W4 becomes 143.6 h), leaving W3 at 136.6 h; cover the remaining 8.6 h with overtime.
- Decline or re-date the extra 20 B unless the shaft subcontract and overtime can be extended - the cell has no room.
Assumptions to verify: standard hours are current and include setup and expected efficiency; the 10% maintenance allowance is genuinely available capacity rather than a reserve; the vendor can hold 15 days on shafts; and W2 is inside the frozen zone, so moving B units requires the authorization that implies. Running at exactly 100% with no queue allowance is fragile - any scrap or downtime breaks it.
3. What does this situation say about the difference between RCCP and CRP, and what planning change would prevent the repeat?
Read the worked answer
RCCP tests whether the MPS is feasible against key work-centre capacity using family-level standard hours. It is fast and coarse by design, and that is exactly where it failed here: it omitted a component that runs on the same constraint cell and it omitted a maintenance shutdown. A rough-cut plan that says 93% in W1 and 88% in W4 while the real shop floor is running shafts and losing a shift is not conservative - it is wrong in the direction that hurts.
CRP, run from the MRP planned orders through the routings, would have caught both: the shaft planned orders carry routing hours on the same work centre, and the maintenance calendar is capacity. The practical sequence is RCCP first to validate the MPS, then CRP after MRP to validate the detailed plan, with the two reconciled - and any work centre that appears in both the MPS family routing and component routings should be flagged as a 'must include' in the rough-cut model.
Prevention: (1) add the shaft hours and the maintenance calendar to the rough-cut model; (2) evaluate capacity inside S&OP before sales commits, not after; (3) hold the frozen zone so the plan is not re-cut weekly; (4) track a capacity-load index per constraint work centre weekly and escalate any week above about 95% before the release, because a plan that is 119% loaded is not a plan.