Supply Chain Variability: what it is, why it's bad & how it can be minimised

Supply Chain variability is best understood by first considering a perfect supply chain with no variability at all: in such a hypothetical supply chain all the material pieces move, or flow, one by one exactly in line with demand; as each piece is consumed by customer demand or a processing work centre it is replaced by the piece behind it. As a result the supply chain holds no static stock and provides perfect service.

When the perfect supply chain starts suffering from variability the movements of the material pieces no longer follow demand and different pieces start moving at different speeds, including coming to a halt. The result is that the supply chain begins to hold static stock, lead-times grow and gaps appear in supply which inevitably lead to capacity losses at processing work centres (when they have nothing to process) and service misses.

Sources of Variability

There are 3 major sources of variability in a supply chain:

1. Batching - inevitable to some extent in most supply chains due to the need for transport/movement economies (eg. raw material supply and shipments) and to enable multi-product processing work centres. Batching obviously leads to static stock, increases lead-times and, if too big, are often absorbing capacity which could be used for processing other more urgently required parts. Lean SMED / batch size reduction and frequent 'milk runs' are the route to reducing the variability caused by batching.
2. Processing issues - these cause materials to stop moving as they await the repair of the processing machine or the correction, if possible, of product quality problems. The result is lost capacity, stock congestion/build up as upstream materials get blocked from progressing, extended lead-times, Planner expedites (see below) and service misses. Lean TPM, TQM, Mistake Proofing, 5S, Standard Work etc are the tools to minimise these sources of variability
3. Forecasts and Planner interventions - supply chains driven by forecasts and master production schedules suffer variability because of forecast inaccuracy - relative to demand most materials are either moving too fast (leading to stock build ups) or too slow (causing service misses) - because, even with excellent 80% forecast mix accuracy, c80% of the items will have >40% forecast error. More significantly, whenever a replenishment schedule is changed by an expedite / re-plan to prevent an imminent backorder, the interrupted materials get stopped, their lead-time is extended and upstream stock catches up and becomes congested. These Planner interventions set off a series of further interventions as the extended lead-times cause further service threats. So, when supply chains are driven by forecasts, master production schedules and expedites, supply chain variability is immense and effectively self-generated. It is easy to estimate the excess of inventory value in a supply chain, due to forecasts and planner interventions, by comparing the actual value of inventory versus the theoretical arrived at by aggregating, for all sku's, their safety stock plus half average order quantity.

The Impact of Variability

Variability has 3 negative effects upon supply chain performance:

1. Inventory - in forecast-driven supply chains growth occurs due to over-forecasting (counter intuitively, under-forecasting also causes inventory to build up (see How under-forecasting generates more inventory than over-forecasting), there is also batch cycle stock (including those queuing to be processed), the holding of safety stock to protect service over the replenishment cycle and stock congestion/build-up due to schedule interventions.
2. Lead-times - grow in line with inventory (Little's Law: Lead-time days = Inventory / daily Throughput), they also fluctuate according to the level of stock congestion and this leads to service threats / misses (because MRP assumes lead-times are known and static) which cause a vicious circle involving further variability generating Planner interventions ->lead-time extensions ->service threats -> more Planner interventions -> etc.
3. Capacity - is lost when schedules are interrupted to enable an expedite and if materials do not arrive when required at a processing work centre due, say, to an earlier schedule intervention. And additional capacity is often released by management as over-time "to clear the backorders" thereby also reducing lead-times and stock congestion. In general the continuous availability, and use, of spare capacity reduces the speed and degree to which lead-times and and inventory grow, but if capacity is tight both lead-times and inventory will grow extremely quickly due to variability.

The below graphic shows how lead-time and inventory grow as variability increases and, as capacity utilisation gets high, they start to grow extremely rapidly. As a result of the latter, at high levels of utilisation lead-times fluctuate wildly in response to variability which also causes major service issues.

On a daily basis the impact of variability on flow can be experienced on roads at rush hour - as the cars move fast (ie. high capacity utilisation) the more prone they are to find themselves in traffic jams (akin to stock congestion and increased lead-time) despite there being no causation accident but due simply to speed variability between the cars; if there are speed restrictions or the hard shoulder is used (both lower capacity utilisation) the jams are far less likely to occur. To see this demonstrated watch the short films at Are all supply planners driving red cars?

How to Reduce Variability and Improve Supply Chain Performance

Despite having practiced S&OP/IBP for many years, companies all too often find themselves providing poor service with excess inventory and frequently having to put on unplanned over-time. S&OP/IBP are useful capacity planning processes (and good for aligning business objectives) but are unable to influence supply chain performance. That can only be improved by reducing supply chain variability.

Lean has been described as being "fundamentally about minimising the costs of buffering variability" (1) and this perspective helps explain the the effectiveness of the many activities that come under the Lean umbrella - SMED enables batch size reduction and TPM/TQM/Poke-Yoke/5S/Standard Work prevents supply hold ups. Lean is effectively about variability reduction (not waste or cost reduction) and by doing so the desired throughput can be achieved using less capacity and inventory so costs are indeed reduced and service levels and cash flow also improve. And these benefits, along with sales growth due to the more competitive supply chain, also increase EBITDA and R/CE.

Now that most manufacturers have been practicing Lean for many years they might feel further improvement is limited - but they are wrong. The most powerful but underused Lean tool is Pull in the form of Enterprise-wide Pull (often also known as Demand Driven MRP) because it virtually eliminates the variability generated by using inaccurate forecasts, master production schedules and Planner interventions to drive replenishment. When implemented through an ERP system, Enterprise-wide Pull consistently delivers:

• Planned service levels, with
• Inventory reductions of c40%
• Lead-time reductions of up to 50%, and
• Cost reductions due to far fewer expedited deliveries / shipments, less over-time and capex avoidance
• All without the need for (impossible to achieve) highly accurate item level forecasts and master production schedules (2)

For more detail on Enterprise-wide Pull, see Factory Flow is non-linear so don't use Master Production Schedules, Neither water or supply chains need Big Tech to tell them how to flow and The Supply Chain Replenishment Problem (and how to solve it)

1 - See To pull or not to pull (Hopp & Spearman 2004) or their book Factory Physics. The former describes Lean as "fundamentally about minimising the cost of variability", the latter describes the Kingman formula (or VUT formula) that formalises the relationships between supply chain variability, inventory, lead-time and capacity as shown in the above graph.

2 - See Demand Driven Institute case studies