A supply chain, industrial operations or distribution operations, is above all a flow. We start with raw materials, and a few steps later we have a finished product that meets a market need.
That “a few steps later” is sometimes an understatement.
I remember a crisis cell when I worked in electronics, because we were short of a specific, innovative component, under allocation. We tracked every manufacturer in our company’s upstream chain – and when we set up this tracking system, we realized that the component in question circumnavigated the globe twice before it had a chance of being delivered to us, and involved numerous players.
This can also be the case at the level of a single company – the flow is complex and not visible. You visit the workshops, we make a routing step on this machine, and then this batch goes… somewhere else. There may be 10, 20, or 30 routing operations, some of which are carried out in-house, others by subcontractors.
The diagram below shows a flow of this type. It extends over several buildings in a factory and includes subcontracted stages.
Imagine yourself standing at the front of one of the machines involved in this flow. You can’t see the flow – you can’t see where it’s coming from and where it’s going. It’s also hard to see if you’re going to meet the customer commitment, the on-time delivery that will happen several steps down the road.
To spice things up, a number of contingencies may arise: a machine may break down, a subcontractor’s lead time may be longer than expected, and a rework operation may have to be added because of a quality problem.
In some industries, it has been possible to put operations in line – this has been a fundamental contribution of Lean manufacturing. That way, you can see the flow from A to Z, and management is simplified.
In many cases, it isn’t possible: we’re stuck with a technological organization, or job-shop organization, with shared resources to support a large spaghetti of flows. This is often the case, for example, in aeronautics, defense, watchmaking…
How do you ensure visibility in this type of flow, so that everyone is working on the right priorities, and that you give yourself every chance of serving the customer on time?
Several techniques can be used.
Implement a steering model
To manage a complex flow, we need to define a few critical points, and a few important milestones, and we’re going to focus the teams’ attention on these points in the flow.
A program will be defined for each of these milestones, and a queue (a time buffer) will be positioned in front of each of these milestones – this makes it possible, on the one hand, to insert safeguards into the flow to absorb contingencies, and, on the other hand, to create a priority management mechanism at these key points. In the case below, we have 4 control points and 4 critical stages that we have selected to control the flow.
Execute according to shared priorities
Each queue is controlled in green/yellow/red. All operations upstream of a control point are prioritized according to their status in the next queue.
In fact, each of these queues makes the previously invisible flow visible to the preceding stages. I feed the critical stage 3 buffer and follow the red/yellow/green priorities of this buffer. The priorities are unambiguous. So is the lead/lag status.
Restricting work-in-progress
The more spaghetti in the dish, the more inextricable and indigestible it becomes.
The more work orders there are in the system, the more inextricable it becomes, and the more priority conflicts there are.
We will therefore only feed the system as fast as it can go. In the case above, it’s step 2 that sets the tempo, and so we only release work orders in step 1 according to the capacity demonstrated in step 2.
View the flow at any time
By collecting data along the entire process, we can now map out where we are in real-time. The progress status of each part can be viewed graphically, delays or advances at key points are visualized on the equivalent of a control tower, and opportunities for improvement are identified through process mining.
Where to start?
Thank you for asking, yes, yes, of course, all these techniques are implemented in Intuiflow… so why don’t we start by getting to know each other?
