Decoupled, demand-driven Supply Chains - faster, more agile & lower cost
Dependent Demand MRP Networks
Ex-stock manufacturers (inc. ATO that hold stock before assembly) with ERP/APS use a fully dependent demand MRP replenishment mechanism upstream from their Master Production Schedule which, in turn, is designed to meet a forecast of independent consumption (be it for customer demand or the assemble to order operation) and a target safety stock. The consequence of this fully dependent network is that every time the system is run it recalculates and recommends changes to the the entire plan up and down the supply chain (DRP, MPS, MRP) to balance back to the inventory targets because
ERP/APS systems are therefore said to suffer from "nervousness" - their plans keep changing, usually very dramatically and often for no obvious reason. Planners know this through the avalanche of MRP exception messages they receive every day that tell them which schedules need expediting or delaying to maintain those safety stock balances - and not helped by the fact that the system provides no advice on prioritisation.
Impact of Schedule Interventions / Expedites by Planners
So Planners usually focus on the expedites to avoid threatened backorders but, in doing so, have no visibility of the additional capacity their unplanned schedule changes are using or the impact of the lead-time increases they're causing on the interrupted schedules.
The impact of both is serious: use of additional unplanned capacity obviously adds to costs and extended lead-times cause materials to arrive late thereby contributing to a vicious circle of yet more schedule changes and lead-time increases - which become particularly severe when capacity utilisation is high (2).
In addition, as lead-times increase so also does inventory (as it's stopped and becomes congested upstream when schedules are interrupted, which is reflected in Little's Law: inventory = lead-time days x daily throughput) so factories and supply chains managed with ERP/APS invariably suffer service issues, growing lead-times, excess inventory and capacity wastage / catch-back capacity costs while Planners/Schedulers are kept very, very busy expediting - the latter of which is effectively disrupting and slowing down material flow.
Blowing aggregated forecasts through a dependent demand network (with an invalid assumption of fixed and known lead-times) for the purposes of capacity planning, as opposed to replenishment, is fine. It provides Planners with a rough idea of capacity requirements but, beware, it also understates them. This is because, when replenishment activity is being driven the same way, the capacity calculation assumes the forecast is 100% accurate so it doesn't capture the capacity required for the unplanned schedule changes and expedites that occur when it proves not to be (3).
If not a Dependent Demand Network then what?
Key criteria are that the methodology should:
This is where decoupling & demand-driven comes in .........
At certain points through the network, planned independent stock locations are positioned and replenished using their own re-order point/cycle mechanisms. As long as the inventory targets &/or re-order points are sized in line with average (plus trend) demand over the lead-time (plus a calculation for its variability and respecting lots/increments) the materials will move, or be pulled, through the supply chain in line with demand without any requirement for damaging interventions & expedites. In addition, item processing at each decoupled work centre can repetitively follow its own optimally efficient sequence and cycle without being actively scheduled or disrupted.
System updates (often daily, but more frequently as desired) do not result in planning nervousness, they simply update the stock position versus the targets (which are periodically re-calibrated - weekly, monthly as desired) and indicate the need, or not, of replenishments that need to be undertaken in the planned sequence/cycle.
Counter Intuitive?
It may seem strange that you are actively parameterising inventories within the supply chain. This is not additional inventory, it is already there waiting to be processed, but through parameterisation it also serves other purposes - it facilitates material flow through pull replenishment, it acts as a cushion, or buffer, to prevent natural process variability being transmitted through the otherwise dependant demand network and it allows the value-add activities to be operated independently of each other and to their own maximally efficient cycle and sequence. Its usage is expected so consumption doesn't generate MRP exception messages and the need for variability generating (4) schedule interventions - thereby enabling material flow and stopping the build-up of, otherwise, excessive and higher levels of unplanned inventory, lead-time & capacity cost (5).
Benefits
Manufacturing supply chains that are Demand-Driven, Decoupled & Sequenced (not scheduled) are, of course, using enterprise-wide Lean pull (using kanban or software supported Demand Driven MRP/S&OP) and the reason they significantly outperform those driven by forecasts, master production schedules and expedites with ERP/APS is that they enable their materials to flow in line with demand. This results in achievement of planned service levels, c40% less aggregate inventory, shorter lead-times, use of significantly less capacity and far more value-add (inc. less expediting) from Planners (6) - all of which means faster, more agile and lower cost throughput.
Suggestions that Lean Pull/Demand Driven MRP supply chains cannot cope with volatile / promotional demand patterns are false: in fact, the more volatile the demand the greater their relative effectiveness v forecast-driven because the latter's forecasts are even less inaccurate & the expediting therefore more copious. When it comes to event management (exceptional promotions, cap. constrained seasonality, NPL, temp.plant shutdowns) Lean Pull/DDMRP, of course, uses forecasts to stock build in advance.
If you're interested in how manufacturers such as yours have benefitted from enterprise-wide pull see Demand Driven MRP/S&OP_Case Studies and for softwares that operate through your ERP transactions system to deliver this replenishment methodology (AND forecast-driven capacity planning/S&OP/IBP plus functionality that recognises & supports event management and other more complex planning challenges) see DDMRP softwares (that can also help you to do a cost/benefit model using your data, &/or run a pilot).
Now it makes sense (?), can you really afford not to?
For the fundamentals see Simple Queueing Theory & how it can help transform Manufacturing & SC performance
also:
NB: buffering/decoupling otherwise dependent demand networks can apply equally to ETO/MTO and project management where the buffers used are that of time and capacity resource: see critical chain project management
I completely agree with you, Simon Eagle: Decoupled, demand-driven supply chains are key to greater agility and cost-efficiency in volatile markets. I see tremendous potential here to overcome the traditional forecasting madness and use resources more intelligently. A 40% reduction in inventory and shorter lead times would be a revolution! Plus, we get exactly what I’m after: more adaptability and flexibility, rather than just more resilience by rigidity.