How to protect your factory from market volatility

Why?

Ask someone involved in manufacturing whether they want factory floor stability and they're likely to ask for clarification:

"What do you mean by stability?"

The answer is:

1. having equipment that doesn't break down
2. having control over the manufacturing process to avoid quality issues
3. having enough capacity/time to do what's expected
4. 100% materials availability
5. predictability - knowing in advance what is next on the schedule and having enough time to prepare
6. working to schedules that are efficient in terms of set up sequencing
7. able to work continuously at a steady rhythm, not being rushed
8. not having schedules change at short notice

Whether you call it 'stability' or something else, people who manage, or work in, manufacturing agree that such an environment is conducive to high levels of productivity, quality and learning (ie. continuous improvement).

Items 1 to 3 are all about standard operations practice and mostly come under the Lean banner - TPM, TQM, SMED, Standard Work, Mistake Proofing, 5S, S&OP. The rest are all impacted by SCM and how material/product replenishment is driven - which is why an adequately stable factory is quite a rarity...........

Why aren't Factories Stable?

Most factories are driven by the output of ERP/MRP/APS systems which use a fully dependent demand network upstream from a single planned independent demand stock location (often customer facing, maybe in front of at ATO operation). These systems are supposedly designed to be responsive to customer orders and every time the plan is calculated the system adjusts planned production schedules to balance back to a target safety stock level. As a result of the dependent demand network, each time the plan is calculated it triggers a change to the entire end to end plan (DRP, MPS, MRP) as actual orders are always different to what was forecast (at the sku level, 80% of forecasts are >40% wrong, even when achieving excellent 80% mix accuracy - 1) or the forecast has changed and / or, because actual production/supply rarely aligns with that planned. The consequence is that these systems end up issuing plans and replans on a frequent basis. Result: the factory is exposed to all the market's volatility (and more because the market volatility gets internally bullwhipped upstream with every plan change) and firefighting is the norm (2).

Firefighting is the absolute antithesis of stability and it actively increases costs because every schedule change uses unplanned capacity, delays the movement of other materials on shared work centres (which then cascades up the routing affecting all other affected items) causing inventory congestion and increasing their lead-times. The latter then pose a service risk (because ERP/MRP/APS assumes lead-times are known and fixed) so yet more expedites and schedule interruptions are encouraged. The result of this vicious circle is on-going wasted capacity, increasing lead-times, excess inventory (through Little's Law: inventory = lead-time days x daily output), continuing service problems and a factory suffering chaos (3).

Can Factories be Stable?

How can SCM provide manufacturing with a requirements plan that allows the factory to be stable while also delivering desired service levels, using right-sized inventories (ie. safety stock + 1/2 ave. OQ for all sku's) and planned capacity (ie. without unplanned change-overs and over-time) ?

The key to this is:

1. Don't use a DRP/MPS/MRP dependent demand supply plan based on inaccurate item level forecasts that is going to need interrupting/expediting to avoid backorders.
2. Don't use a replenishment methodology that keeps re-planning everything

Instead of incorrectly telling the factory how much to buy, make and ship of what & when, ask them to operate using simple re-order point/cycle pull mechanisms. These allow materials to start moving from suppliers and through the factory and distribution in line with actual demand (instead of an inaccurate forecast) and don't use disruptive replans.

ROP/ROC mechanisms can be designed for each item at each planned stock position (eg. finished goods network, factory, bulk break points within the factory, raw materials) to reflect future average demand/consumption, the desired service level (given the level of demand variability), the length of the lead-time, the desired frequency of supply and respecting MOQs/increments.

SKU demand patterns nearly always have a stable and accurately forecastable average (its the 'noise' that causes weekly/monthly point forecasts to keep changing and be 'inaccurate') and its this that is used to calculate the 'order-up-to' stock target (to cover the lead-time, the MOQ or order cycle and inc. something calculated for variability error) which, when replenished to with the planned frequency, delivers the desired service level.

Sometimes, of course, demand has trend so the target needs to be periodically calibrated to reflect this and sometimes there are abnormally large demands (ie. outliers from a normal distribution) and these need managing as exceptional events (eg. anticipated via collaboration with the sales team) and built for in advance, supplied on future capacity feasible dates or supplied immediately with knowledge that a stockout will likely occur.

And as well as accommodating desired replenishment frequencies/quantities, ROP/ROC mechanisms allow manufacturing work centres to adhere to stable and efficient replenishment sequences without capacity wasting interruptions.

Agility thru' Stability?

So yes, factories can indeed be 'agile' (ie. consistently deliver planned service levels) while also being stable and in doing so they use significantly less inventory (usually c40% less) and capacity (ie. lower costs) and shorter lead-times than those that try to be 'responsive' using traditional ERP/MRP/APS and their associated re-plans, expedites and schedule interventions.

Allow materials to flow in line with demand using forward looking ROP/ROC pull mechanisms; use them also for de-coupling the BOM to prevent the propagation of process variability and only make inventory in advance of orders when essential (eg. significant promotions, big new customers, capacity constrained seasonality, NPL, plant holiday shutdowns).

How?

All of this can be achieved using your legacy ERP with a new breed of enterprise-wide pull softwares that support the entire process - where to de-couple if its not obvious, sizing and calibrating of ROP/ROC inventory targets, recognition & management of abnormal demand/forecast events, generation of recommended replenishment orders, provision of 'low' inventory/service alerts, S&OP/work centre capacity planning and many other 'off the shelf' functionalities to manage the planning complexities (eg. container/shipment fill, co-make items, shelf life/batch # expiry management etc) found in particular industries.

For a deeper understanding see Factory flow is non-linear so don't use master production schedules and www.demanddriveninstitute.com for software providers, case-studies, academic research, education, conferences etc.

1 - forecast mix accuracy = 100% - (aggregate of forecast errors / aggregate of forecast)

2 - time fences are supposed to protect the factory from disruption but often lead to stock outs so get over ruled or lead to a system generated backlog of unmet net demand

3 - the destructive impact of fire-fighting generated variability upon supply chain/factory performance is explained by simple queuing theory here and for how variability impacts supply chains see this