Skip to content
Opt Dir

Manufacturing ยท Aggregate Production Planning (APP / HMMS)

How Do I Balance Production Rate, Inventory, Overtime and Subcontracting Over the Next 12 Months?

Manufacturing 4 min read
#aggregate production planning #medium term planning #workforce planning #overtime #subcontracting #hmms #app #sales and operations planning

Mid-size manufacturer's problem of jointly planning production rate, inventory, workforce, hiring and firing, overtime, subcontracting and backlog across multiple products and 3-18 months. Known in OR as Aggregate Production Planning (APP); canonical formulation by Holt-Modigliani-Muth-Simon (1960).

In plain words

You run a mid-size manufacturer with 10-50M TRY monthly revenue, 5-30 product families and 30-200 workers; sales hands you a one-year demand forecast where peaks and troughs vary sharply month to month. Each month for each product family you must make five interlocking calls at once: run at a steady rate and build inventory, hire and fire workers, cover peaks with overtime, subcontract to sub-tier manufacturers, or push orders into next month. Get the mix wrong and average inventory bloats 20-40%, overtime breaks the 270 hours per year legal cap, last-minute subcontracting closes 20-40% above market price, and late deliveries lose customers. Splitting these five decisions across sales, production and HR with three separate intuitions burns 4-25M TRY of operating margin each year.

Sound familiar?

  • We are a 10-50M TRY monthly-revenue mid-size manufacturer; 5-30 product families, multi-product line of 30-200 workers; sales gives a yearly demand forecast but turning that into a monthly production rate stays scattered in the planner's head.
  • Demand is seasonal โ€” peak in summer, trough in winter, or the other way around; the question 'steady production into stock or month-to-month rate change?' is re-litigated each year from scratch.
  • Hire-fire and temporary-contract decisions are taken by HR, production and finance based on three separate gut feels; collective-agreement constraints (Turkish Labour Law, 270-hour annual overtime cap) enter the plan only afterwards.
  • Overtime, second shift, or subcontract โ€” every product family decides 'this time' on the spot; nobody calculates which yearly mix is cheapest.
  • The trade-off between backlog (order deferral) and inventory is intuitive; the amount we defer with 'next month' to customers is not a policy, it is an ad-hoc call.
  • Subcontracted work fluctuates 5-25% year over year; the capacity ceiling (the hours a subcontractor can give us) is not recorded in the annual plan.
  • We had tackled lot sizing (problem 16, single-product 'should I produce this month, how much?'); APP sits above it โ€” workforce and subcontracting decisions belong here, but the two live in different software modules.

Why it matters

Losses from intuitive medium-term production planning: (1) stock-shift-overtime mix mis-calibrated โ€” average stock is 20-40% higher than needed, capital tied up at 20-35% yearly carrying cost, (2) unnecessary hire-fire churn โ€” temporary-worker hiring, training, severance and morale loss runs 5-25 thousand TRY per worker, (3) mandatory-overtime spike โ€” overtime premium (50% weekday, 100% weekend) consumes 15-25% of yearly capacity, fatigue and quality slippage add further cost, (4) last-minute subcontracting โ€” unplanned subcontract is closed at 20-40% above market rate, (5) intuitive backlog โ€” customer-relationship penalty and contract delivery fines run 2-5% of annual sales. Field evidence shows numerical medium-term planning cuts total production+workforce+inventory cost 8-20% and reduces hire-fire churn 30-60%. For a 100-300M TRY annual-revenue mid-size manufacturer, the band corresponds to 4-25M TRY/year additional operating margin.

How it's solved

Technical depth

In one sentence: Lay 12 months across a spreadsheet, solve for the production-rate + inventory + headcount + overtime + subcontract mix that minimises total cost for each month, and refresh the answer monthly as the forecast updates โ€” a numerical scale replacing intuition.

In Operations Research (the discipline that uses math and computers to solve business decisions) literature this is Aggregate Production Planning (APP) โ€” multi-product, multi-period (typically 3-18 months) joint capacity and workforce planning at the medium-term horizon. The canonical formulation is the HMMS model from the 1960s. Three-stage solution:

1. Modeling. Decision variables per period (month): per-family production quantity, end-of-period inventory, backlog (deferred orders), regular-time production, overtime production, subcontract quantity, end-of-period workforce, hires, fires. Inputs: per-family per-period demand forecast, hours-per-unit conversion, regular labour-hour cost, overtime hour cost (TR Labour Law 50% premium), hire cost (sourcing, training, onboarding), fire cost (severance, notice), inventory holding cost (% capital + warehouse + insurance), backlog penalty (contract fines + customer-relationship), subcontract cost, capacity ceilings (regular hours, overtime hours โ€” TR Labour Law article 41, 270-hour annual cap per worker, subcontract ceiling). Constraints: inventory balance (prior stock + production + subcontract โˆ’ demand = end stock โˆ’ backlog), workforce balance (this period = prior + hires โˆ’ fires), hour capacity (hours required โ‰ค regular + overtime), inventory capacity, overtime annual cap, subcontract ceiling, collective-agreement hire-fire bounds. Objective: total cost minimum across the horizon โ€” regular wages + overtime premium + hire + fire + holding + backlog penalty + subcontract.

2. Solver-driven decision. Three classical approaches: (a) HMMS QP (Quadratic Programming โ€” squared cost penalties yield a closed-form decision rule; smooths swings in workforce and production rate). (b) LP/MIP โ€” linear unit costs handled by LP (Linear Programming โ€” minimise a linear objective under linear constraints), integer hire-fire or overtime via MIP (Mixed-Integer Linear Programming โ€” optimisation with some 0/1 variables and some continuous). Mid-scale (10-30 product families ร— 12-18 periods) solved by commercial or mature open-source MIP solvers in minutes. (c) Scenario-based (stochastic) APP โ€” under demand uncertainty, multi-stage stochastic programming or robust optimisation; alternatively, deterministic approximations adapted to uncertainty. (d) Heuristic โ€” search decision rules (regression on the planner’s past decisions to derive a rule). Rolling-horizon (re-solve monthly) re-planning: first 1-2 months “frozen” (committed to contracts), 3-12 months “flexible”.

3. Field integration. Output in three layers: (i) monthly production-rate + inventory + workforce + overtime + subcontract decision table (feeds S&OP meeting), (ii) monthly product-family disaggregation into weekly lot sizing (problem 16), (iii) finance + HR projection (cash flow, payroll projection, hire-fire calendar). Data flow: sales forecast (CRM + historical), capacity inventory (machine-hours, workforce), collective-agreement and Labour Law constraints, subcontractor capacity confirmations. Monthly S&OP committee: sales-production-finance alignment, plan-actual gap, demand-forecast refresh, overtime-subcontract mix recalibration.

Alternatives

Manual plus spreadsheet plus production manager's experience

Free

Zero license

Who it fits: Small manufacturer (1-3 product families, 20-50 workers), single-shift line

  • + Zero software cost
  • + Flexible, accommodates last-minute changes
  • + Good outcome with an experienced production manager
  • โˆ’ Mental capacity insufficient above 5 product families
  • โˆ’ Hire-fire / overtime / subcontract mix is intuitive, no optimality
  • โˆ’ Stochastic demand and Labour Law constraints get missed
  • โˆ’ Sales-production-finance alignment at S&OP is weak

ERP production planning module

Enterprise

150K-600K TRY license + 30K-90K TRY/year maintenance (TR market observation)

Who it fits: Mid-size manufacturer, 5-30 product families, multi-shift

  • + ERP-integrated โ€” sales, stock, production on one system
  • + Local-language UI, local support
  • + MRP plus rough-cut capacity included
  • โˆ’ Medium-term planning optimisation module usually absent โ€” plan done by hand
  • โˆ’ Hire-fire and overtime as decision variables out of scope
  • โˆ’ Uncertainty-aware planning support thin

Advanced Planning System (APS)

Enterprise

300K-2M EUR license + 60K-300K EUR/year maintenance (international solution providers)

Who it fits: Large manufacturer, multi-site, 30+ product families, capacitated line

  • + Exact-programming or quadratic-smoothing-based medium-term planning is mature
  • + Sales-operations planning, material-requirements planning, capacitated lot-sizing integrated
  • + What-if scenarios, rolling horizon
  • โˆ’ High license plus 12-24 month deployment
  • โˆ’ TR Labour Law and collective-agreement customisation adds project time
  • โˆ’ Wide training programme for the operations team

Open-source solver plus custom APP module

Open Source

License free; in-house build 16-30 weeks or 600K-1.8M TRY consultancy

Who it fits: Manufacturer with a tech team, where integration with the existing ERP is required

  • + No license fee
  • + Linear and quadratic medium-term planning methods are well-defined for open-source solvers
  • + Uncertainty-aware and rolling-horizon extensions exist in open literature
  • โˆ’ Requires in-house solver specialist plus enterprise-resource-planning integration team
  • โˆ’ Moving from prototype to field system takes time
  • โˆ’ Maintenance burden stays with the operator

Recommendation

Small
1-3 product families, 20-50 workers: stay on manual plus spreadsheet. Three core rules (pre-parameterise the stock-overtime-subcontract mix from peak-to-trough demand spread, freeze hire-fire on a quarterly basis, track 270 yearly overtime hours per worker) yields 5-10%. Optimisation software will not pay back.
Medium
5-30 product families, 50-300 workers: enterprise-resource-planning production module plus custom medium-term-planning add-on, or a compact advanced-planning-system optimisation module. 6-9 month pilot. Expected 8-15% total cost, 30-50% hire-fire churn, 10-20% overtime spend improvement. Payback 18-30 months.
Large
30+ product families, 300+ workers, multi-site: full advanced planning system + medium-term planning + material-requirements planning + sales-operations planning integration. 1-3M EUR/year total investment. Payback 24-48 months. Total production+workforce+inventory cost 12-20%, and noticeable plan-stability gains under demand uncertainty, are typical.

Ask in the meeting

  • What method is used for medium-term production planning โ€” quadratic-cost smoothing model, linear / integer mathematical programming, heuristic decision rule, or only rough-cut capacity visualisation?
  • Does the model jointly optimise production rate + inventory + workforce + hire-fire + overtime + subcontract + deferred orders across multiple product families ร— 12-18 periods, or is it layered?
  • How are TR Labour Law constraints (270 annual overtime hours per worker, 50% weekday and 100% weekend premium, severance) and collective-agreement hire-fire bounds defined in the system?
  • Is scenario-based planning supported for demand uncertainty, or only single-forecast (deterministic) plans?
  • How is the rolling-horizon structure configured โ€” first 1-2 months 'frozen' (committed), the rest 'flexible'?
  • Is disaggregation from monthly product-family levels to weekly batch-size decisions automatic, or a separate module?
  • In a 12-16 week pilot with real demand and capacity data, what kind of savings report can be produced against the prior manual plan?
  • If we end the contract, in which standard format can we export demand-forecast history, production-plan archive, hire-fire calendar and overtime-subcontract data?

Technical details

Editor’s note

Colloquially this is the “annual production plan”, “medium-term plan” or “S&OP” (Sales & Operations Planning โ€” the monthly sales-and-operations alignment meeting). Academically it is Aggregate Production Planning (APP), with the canonical HMMS model from the 1960s. A single LP/QP captures production rate, inventory, workforce (hire-fire), overtime and backlog as continuous variables and trades off the cost of inventory build-up against regular capacity, overtime premium and hire-fire churn.

Do not confuse this with problem 16 (Wagner-Whitin lot sizing). Lot sizing answers, for a single product, ‘should I produce this month, and how much?’ โ€” a single-layer trade-off between setup cost and inventory holding. APP sits above: it jointly decides production, workforce, overtime and subcontracting across multiple products on a 3-18 month horizon. Solved together they produce an internally consistent plan: APP first sets the monthly production rate and the workforce-overtime-subcontract mix at the product-family level, then lot sizing breaks the monthly quantity into weekly batch sizes. This hierarchy is the standard framework in production-planning textbooks.

Do not confuse it with problem 1 (job shop scheduling) either. JSP is a daily operation-sequencing schedule; APP is far above that, on a monthly-yearly horizon, planning the mix.

Most-skipped point in the sector: rolling-horizon discipline between forecast and plan. Academic APP models typically solve the full 12-month horizon at once; in real production the first 1-2 months are committed (“frozen”), the next 3-12 are “flexible” โ€” every month the forecast is refreshed and the plan re-solved. Without the frozen window the plan changes from scratch every month and the production team loses confidence. Second most-skipped: the 270-hour annual overtime cap under TR Labour Law article 41 (per worker). This must be a hard constraint in the APP model; otherwise the plan says “we can always solve it with overtime” and crosses the legal limit, triggering penalties and union pushback.

Step-by-step path โ€” for the SMB

Stage 1 โ€” Measure first, plan second. At least 12 months of data: per-family monthly actual production, actual sales, end-of-month stock, hire-fire movements, overtime hours, subcontract quantity, backlog (late delivery). Log the demand forecast: compute forecast-vs-actual error (MAPE). Capacity inventory: hourly capacity per line (regular + overtime), workforce skill matrix, subcontract ceiling (per subcontractor).

Stage 2 โ€” Extract the knowledge asset. Hours-per-unit conversion coefficients, unit costs (regular, overtime, hire, fire, holding, backlog penalty, subcontract). Constraints: warehouse capacity, 270-hour annual overtime cap, collective-agreement hire-fire bounds, contracted-customer delivery penalties.

Stage 3 โ€” Pilot. 12-16 weeks. For a subset (e.g. 5-10 main product families) run the APP MIP plan in parallel with the manual plan. Decision stays with the planner; APP recommends. Success criteria set in advance: total production+workforce+inventory cost -8% minimum, hire-fire churn -25% minimum, zero breach of overtime cap.

Stage 4 โ€” Rollout. 9-15 months to all product families + lot-sizing disaggregation + S&OP-meeting output integration. Monthly rolling-horizon re-planning. Monthly S&OP committee: sales-production-finance alignment, plan-actual variance, forecast-MAPE refresh, overtime-subcontract mix recalibration.

Risks โ€” what can go wrong

  1. Demand-forecast slip. APP takes the deterministic forecast as input; if MAPE exceeds 20-30% the plan drifts from month one. Stochastic / scenario-based APP, or rolling-horizon monthly re-planning, is required.
  2. Collective-agreement constraint change. A mid-year collective-agreement renewal alters wages, overtime rules and hire-fire bounds; APP parameters must be recalibrated semi-annually with an emergency re-plan after the signing date.
  3. Currency (FX) cost. Imported raw material and equipment cost moves with TRY-EUR/USD; if unit production cost is treated as constant, FX jumps invalidate the plan. FX-scenario APP, or monthly parameter refresh, is essential.
  4. Single-supplier lock-in. A single subcontractor means the capacity ceiling may not materialise (subcontractor takes other work, capacity closes), and the APP plan collapses. At least 2-3 subcontract sources plus annual framework agreements are needed.

Solution method โ€” technical view

ApproachTypical scaleSolve timeGuaranteed optimum?
Heuristic (production manager + rule)1-5 product families, 6-12 monthsinstantNo, 50-70% optimum
HMMS QP (quadratic, smooth)5-15 product families, 12-18 monthsminutesYes (closed-form or QP)
Transportation-tableau LP5-20 product families, 6-18 monthsminutesYes (LP)
MIP (integer hire-fire, integer overtime)10-30 product families, 12-18 monthsminutes-hoursYes (within bound)
Stochastic / scenario-based APP10-30 product families, multi-scenariohoursYes (LP/MIP), bounded scenarios
Search decision rules (regression on planner’s past decisions)Statistical ruleinstantNo, calibrated to experience
Rolling-horizon re-planningMonthly re-solveminutes/monthYes (each solve)

Objective function choices:

  • Objective 1 โ€” Total cost minimum: Regular + overtime + hire + fire + holding + backlog + subcontract. Standard APP objective.
  • Objective 2 โ€” Service level maximum: Drive backlog to zero, contract penalties first.
  • Objective 3 โ€” Production-rate smoothing: Minimise oscillation on workforce and machine (smooth operations).
  • Objective 4 โ€” Hire-fire churn minimum: Employer-brand and morale priority.

Multi-objective: weighted sum (total cost + churn weight) or hierarchical (zero contract penalty first, then total cost minimum).

Academic sources

Listed in the page frontmatter under sources.

Sources

  • Holt, C. C., Modigliani, F., Muth, J. F. and Simon, H. A. (1960). Planning Production, Inventories, and Work Force. Prentice-Hall. Foundational APP reference, HMMS model.
  • Nam, S. J. and Logendran, R. (1992). Aggregate production planning - A survey of models and methodologies. European Journal of Operational Research, 61(3), 255-272. Comprehensive survey.
  • Mula, J., Poler, R., Garcรญa-Sabater, J. P. and Lario, F. C. (2006). Models for production planning under uncertainty: A review. International Journal of Production Economics, 103(1), 271-285. Production planning under uncertainty review.
  • Bitran, G. R. and Yanasse, H. H. (1984). Deterministic approximations to stochastic production problems. Operations Research, 32(5), 999-1018. Deterministic approximations to stochastic production problems.
  • Vollmann, T. E., Berry, W. L., Whybark, D. C. and Jacobs, F. R. (2005). Manufacturing Planning and Control for Supply Chain Management (5th ed.). McGraw-Hill. Textbook framework for production planning and control.
  • Yร–K Thesis Center โ€” keywords: ’toplu รผretim planฤฑ’ or ‘รผretim planlamasฤฑ’ or ‘kapasite planlama’ โ€” 35+ theses. tez.yok.gov.tr

Glossary

Aggregate Production Planning
Medium-term OR/MS problem of jointly setting production rate, inventory, regular workforce, hiring/firing, overtime, subcontracting and backlog across a multi-product, multi-period horizon (typically 3-18 months) to minimise total cost.
Workforce Capacity Balance
Trade-off between regular workforce (smooth but expensive in idle months), hiring/firing (cheap month-to-month but high churn cost), overtime (capacity flex but premium pay plus fatigue) and subcontracting (flex but quality and lead-time risk). Captured as decision variables in APP.
MIP
An optimization model where some decision variables are forced to be whole numbers (e.g. number of trucks, number of shifts).
Was this helpful?
Suggest correction

Related problems

A New Part Order Arrives โ€” How Do I Get the Optimal Operation Sequence + Machine Choice + Setup?

If you are an SMB CNC manufacturer, tooling shop or engineering workshop producing 50-500 different parts, every new order puts the same decision in front of you: you take the customer's CAD model and have to work out on which machine, in which order, with which tool and fixture, and in how many setups the part will be made. If you simply write down the first feasible sequence that comes to mind, setup time grows 3-5x and parts that miss tolerance are reworked; finding the right sequence means comparing several alternative routings for the same part. When the decision lives only in one engineer's head, the similar-part memory walks out of the door when that engineer leaves, and routings of older parts are not refreshed when a new machine is bought. This page is for production-engineering teams who want to make the operation sequence and the machine assignment written and comparable as the bridge from design to make.

Manufacturing 7 min

Backward from the Customer Order โ€” Which Raw Material, When, and How Much Should I Order?

For mid-size manufacturing SMBs running 50-500 end items and 200-2,000 raw materials and sub-components (automotive Tier-2, white-goods component shops, furniture and assembly, machinery). With each product's parts list (BOM) and each part's procurement or production lead time on file, the question is: a customer orders 200 units of product X with a fixed delivery week โ€” which raw materials in what quantity must be ordered which week, and which sub-parts must enter assembly when? Material requirements planning (academic name MRP) answers this question backwards from the delivery date and is the core logic underneath every ERP. Done on paper or by 'my supplier always takes 3 weeks' rules of thumb, you either run out and stop the line, or burn cash on excess stock.

Manufacturing 8 min

Five Suppliers Quote the Same Part โ€” How Much Do I Buy from Each?

You run a manufacturer with 20-200 staff and buy the same raw material or component from 4-8 different suppliers. Each supplier differs on price, quality, lead time, capacity, payment terms, and financial stability; deciding who stays on the approved list and how much of each order goes to whom are two separate calls. A wrong pick can trigger a months-long quality crisis or stop the line when a single supplier fails; the 'cheapest bid wins' approach hides quality, late-delivery, and compliance costs and ends up 15-40% more expensive. Scoring 5-15 bids across 8-12 criteria by hand in a spreadsheet becomes inconsistent fast.

Manufacturing 4 min
Esc Close