Line balancing — Or how you can trap yourself with best intentions to not waste resources

TL;DR Balancing the capacities of a line tries to avoid any excess capacity, effectively making each capacity matching the bottleneck’s. To protect against flow disruptions we need to buffer each. That increases inventory and lead time compared to drum-buffer-rope systems with untouched capacities. With a balanced line we loose the ability to boost flow time of individual work items by using excess capacity to allow more setups to reduce batch sizes. But more prominently we loose the flexibility of system to be able to respond to changes in market demand and composition (product mix).

Assumptions:

• Any work package going through different (linear) working steps requires different time at each.
• Since you can’t buy half a machine or hire half a person for a shift, capacity number is discrete, meaning it increases in jumps that are not in the chunks you actually need them.
• We have different resource requirements for the work at the different resources while we have different capacities of them.
• As a result, we usually have unbalanced capacities in a manufacturing, project or mixed environment.

The natural reason why a bottleneck emerges

That, in fact, is the reason, why any system naturally has a resource that can put through less work packages than others.

Let’s look at it alone and ask: “Can it process the total number of work packages that are due within the given time frame?”

If it can’t accomplish that, it’s the bottleneck of the system. We put a higher demand of it than it is capable of.

Usual approach to manage capacities

• How do we usually approach managing capacities? They in fact cost us money, and we don’t want to waste money.
• Therefore we conclude that using their capacity most times, we make use of the money they cost us.
• Now, if we do that in a linear system with a bottleneck, we pile up excessive inventory in front of bottleneck.
• If that’d disappear, we would have a significant reduction of lead time; because there were no queues.

What is line balancing?

So, one actual way to achieve this is to resize all capacities of the system to have the same throughput at full utilisation. You sell/buy redesign capacities so the throughput of each is the same. After you’ve done this major effort, you have a system where you don’t waste resources. We could label that as a perfect lean operation.

On the other hand, we make each resource a bottleneck. If any resource is starved because of upstream disruptions, we loose overall system throughput. That’d require us to introduce time buffers in front of each resource. That would worsen our lead time.

But still, we might be better off in lead time compared to before. We might.

Usually any company is producing more than a single product. Even if we know the product mix roughly from experience, quite often we exhibit natural fluctuations within the product mix. You might be able to forecast category demand quite well, but how great can you forecast exact product mix? Put it in the AI stuff. How can and will it be able to determine which customer is (not) coming around the corner in the next minute?

As we recall that different products require different resource capacities, we would no langer have a balanced line. We would waste capacities again. We would need to constantly rebalance the line, which is cumbersome. We might actually regret reducing some capacities that we would need right now but can’t resize as fast as product mix changes! The same applies to changing market demand: Which market has a constant demand you can rely on getting your constant share of? If marked demand changes, how fast can I resize the whole balanced line to match that? And when to do it and when not to do it? The system becomes really inflexible whereas we actually require more flexibility today given rapidly changing markets.

What do we actually want?

We actually want flexibility built in the system to be able to respond to markets quickly.

We still want reduced lead time and less inventory in the system (which is the cause of the former).

We want to protect the system’s throughput against upstream disruptions. And we surely don’t want to put buffers everywhere.

Why is having a bottleneck great?

Actually, it’s great if we just have one bottleneck! It’s an asset.

That enables us to put the time buffer in front of it (apart from shipping).

Why is having excess capacity at most resources is an asset?

Bottleneck limits system’s throughput. So we don’t want to waste setups there. So we might want to reduce setups at bottleneck by batching alike work. We do that because the capacity is consumed and we want to increase the effective progress done.

But what is the state with non-bottleneck resources?

As they are no bottleneck, we don’t use all of their capacity.

Do we want to save on setups here? If we did, we would have more excess capacity, which has no value to us.

What if the excess capacity is actually an asset?

Since we have excess capacity, we can deliberately increase the setups to reduce the batch size.

We make the individual work item available faster to the bottleneck.

This reduces flow time of the individual item.

How does bottleneck system is affected by fluctuations in demand or product mix?

If the demand changes, nothing changes since the bottleneck. Compare this to the balanced line, where you need to increase the capacity of the whole line!

If really push comes to shove and (after offloading work from bottleneck that also others can do, reducing number of setups, reducing time per setup) you increase capacity of bottleneck resource, until the bottleneck shifts. Up and until then, you only need to invest in increasing the capacity of the single resource. That is both faster and cheaper.

Changing product mix means resource requirement changes, again, excess capacities might turn out to be really valuable. They become faster valuable than you can increase them with a balanced line. Of course, you need to consider the time to figure it out, too.

With changing product mix the bottleneck might move, too.

How then do we reduce excess inventory in a non-balanced line?

Since the bottleneck limits overall system throughput, piling up inventory happens only if we release more work to the system than it can handle. So, if we only release new work to the system when the bottleneck has just finished one, there is no bulid-up of inventory.

Of course, we protect it from disruptions upstream with a time buffer in front of it.

Recap:

• A balanced line assumes non-bottlenecks are to be operated like bottlenecks — under this very misassumption a balanced line actually creates a problem that previously didn’t exist
• Once all capacities have no excess capacity, it turns each into a bottleneck
• Therefore it is very prone to disruptions anywhere in the work flow
• Protecting against this requires buffers everywhere which increases lead time again
• We can no longer let work items flow faster from upstream steps to the bottleneck by reducing batch sizes by allowing more setups at resources with excess capacity; a balanced time once more means reduced flow time, thus longer lead time.
• It is very hard to maintain if market demand changes in volume and composition (product mix).
• To be able to tap a changing market demand in its volume and composition (product mix) requires again hard work to resize the system’s capacities, which customer will wait in today’s time “wait, I can serve you, until I’ve increased the capacity of my line?”

Assess the state of your work flow in project and mixed environments (like engineer-to-order)

Register for the Dolphin Study to assess (1) the scale of your problems by assessing the severity of multiple root causes (benchmark with other participants) and (2) a first indication of your true potential of your work flow. It’s free and just a couple of questions you are bet to be able to answer!

Here is the link: https://dolphinstudy.info