Risk Analysis for Manufacturing | Vose Software

Industries / Manufacturing

Is the capacity actually enough?

Model plant capacity, Six Sigma quality and supply chains under uncertainty.

  • ModelRisk
  • ModelChoice
  • Pelican

Five-year cost of keeping four presses

10,000 futures

P80 $1,584k
80%within20%above
Overtime, subcontracting and lost sales over five years, US$ thousand
80% within20% above
From the free example model MF-07 Capacity Expansion Decision, no new press. 10,000 futures.

The plan's $426k holds in 42.2% of futures. Simulated, the mean is $920k, and one new press saves $156k on average.

Clients in manufacturing

  • Dow
  • DuPont
  • Air Liquide
  • ZF
  • Eaton
  • Emerson
  • Olympus
  • Duracell
  • Chemours
  • LANXESS

The lifecycle

Where uncertainty enters a plant

From the drawing to the warranty claim, each stage has its own question. Pick one to jump to it.

Questions we answer

The questions your board asks, with the numbers behind them

Charts marked with a model code come from a free example model you can open in Excel. The others are marked illustrative.

Where each product fits:
1DesignModelRisk

Will the assembly stay within tolerance?

Worst-case tolerancing says the gap can close completely; root-sum-square says 1,545 parts per million out of specification. Simulating the parts as they are really made gives 8,799, and shows which two changes bring it to 1,400.

Root-sum-square, every part centred1,545 ppm
Simulated, parts as made today8,799 ppm
Capable bearing supply3,700 ppm
Capable bearing, centred spacer1,400 ppm
From the free example model MF-02 Tolerance Stack-Up and Process Capability: parts per million out of specification. 10,000 futures.
2ProductionModelRisk

Will a run of 5,000 meet the order?

Every part costs $40 to make and every part short costs $250. A run of 5,000 has the lowest expected cost. If the 3% defect rate were known exactly it would always meet the order; known only from 400 inspected parts, it meets it in 80.8% of futures.

Order 4,800
19.2%short80.8%order met
Good parts from a run of 5,000
19.2% short80.8% order met

Meets the order 80.8% · expected cost $201.5k

From the free example model MF-01 Production Yield and Scrap. 10,000 futures.
2ProductionModelRisk

Is a standby machine worth its cost?

Four machines run in series, so a stop anywhere stops the line. The capper causes 41% of the downtime; a standby capper cuts it from 103.7 to 66.1 hours a year on average and pays for itself in 83.8% of years.

Line downtime in a year, hours

Mean 103.7 h · availability 98.27%

From the free example model MF-03 Reliability and Availability of a Production Line. 10,000 futures.
3SupplyModelRisk

Which protection against supplier disruption pays?

Treat each supplier's outages as random events of random length, including a regional disaster that stops two suppliers at once. Thirty days of stock gives the lowest average cost; insurance gives the smallest 1-in-100 year.

Do nothing$3.25M
More safety stock$1.95M
Second electronics supplier$2.56M
Business-interruption insurance$3.66M
From the free example model MF-06 Supply-Chain Disruption: the average yearly cost of disruption. 10,000 futures.
3SupplyModelChoice

Dual source, stay offshore, or make it ourselves?

A consumer electronics maker weighs its sourcing after recent global disruptions. Staying with the offshore supplier, with better audits and buffer stock, has the best expected value, $12.1M; building its own factory comes out at $2.3M.

Nearshore + Dual SourceEV $10.7MSingle Offshore SupplierEV $12.1MBestVertical IntegrationEV $2.3MStatus QuoEV $10.4M

From the free example Supply Chain Resilience Decision, rolled back in ModelChoice.
4After salesModelRisk

How much should we provide for warranty?

Nobody knows yet whether the redesigned drill fails less than the old one. The year's warranty cost is $69.3k on average and $88.9k in a 1-in-20 year, and the year-end reserve needs $38.8k on average.

Expected $69.3k
52%within48%above
Warranty cost of the year's sales, US$ thousand
52% within48% above
From the free example model MF-05 Warranty Cost Forecast. 10,000 futures.
+Every plantPelican

What risks are we carrying across every plant?

One shared register for every plant, with owners and actions. It takes in results from your ModelRisk models, so the register and the numbers stay in step.

SiteRiskOwnerBeforeAfterStatus
Plant A, castingsSingle foundry in a typhoon regionSupply chain lead209Mitigating
Plant B, assemblyCapper failures stop the lineMaintenance manager166Mitigating
Plant C, mouldingDemand outgrows four pressesOperations director158Open
GroupWarranty claims on the new drillQuality manager126Open
Plant A, castingsEnergy price spikeFinance controller124Mitigating
Illustrative A group register with scores before and after mitigation (5 × 5 scale).

How it fits together

One risk picture, from the plant register to the numbers

The same risks feed the models and the decisions, and the results go back to the register.

  1. Step 1 · PelicanCapture the risksSupplier, plant and product risks with owners and actions, site by site.
  2. Step 2 · ModelRiskQuantify themYield, downtime, disruption and warranty ranges give the cost at each confidence level.
  3. Step 3 · ModelChoiceChoose the responseLay out the protections with their risks and roll the tree back.
  4. Step 4 · PelicanReport backResults return to the register for plant reviews and the group view.

Training

Risk analysis training

Risk analysis modelling courses can focus on your industry and on the problems you model, online or face-to-face.

See the courses

Consulting

Bring one model

Our consultants review one of your own models before a critical decision or a public review.

See consulting

FAQ

Common questions

Why not use root-sum-square tolerancing?

Because it assumes every part is centred and normally distributed. In the free tolerance example, root-sum-square says 1,545 parts per million out of specification; simulating the parts as they are really made gives 8,799.

How much safety stock do we need?

Enough for the demand during the lead time as it really is, including the delayed orders. In the free inventory example the Normal formula, aiming for 95% of cycles without a stock-out, delivers 91.5%.

Do we have to leave Excel?

No. ModelRisk and ModelChoice are Excel add-ins: you add ranges, risk events and decisions to the workbooks you already have.

Can we try it on one of our own models?

Yes. ModelRisk and ModelChoice each have a free 15-day trial, and our team can go through one of your own models with you. For Pelican, we will show it to you on risks like yours.

Bring one model. We'll show you what drives it.

Free 15-day trials of ModelRisk, ModelChoice and Tamara. For Pelican, we'll show it to you on risks like yours.