Buffers drain the laundry’s lifeblood
In the industrial laundries buffers are everywhere. But they come at a cost.
By definition material flow capacity is determined by process route capacities. In turn process route capacity depends on batch and category sequences. We know that from the previous blog entry. And since most changes in batch and category sequences require setting up, all times, not only efficient machine times, become important – also the time spent on preparations before a machine starts, as well as after it finishes processing.
The significance of repetition
The linen and textiles must be ready – sorted out by categories, weighed, positioned and ready to be loaded – when a washer extractor ends a wash program, being manual preparations all of them.
But what is more important, and has a much greater impact because of its frequency and systematism, is the time mechanical preparations delay or postpone the processes.
With increasing frequency even the shortest mechanical preparations become important. The highest frequencies – or rather the shortest cycle times – in the laundry are usually found on the continuous batch washers.
Right down to every 120th second batches are replaced in the compartments. During an 8-hour shift that amounts to some 200 cycles. If for instance 10 seconds are wasted in each cycle, or if processes can be improved in such a way that the efficient processing time could be reduced by 10 seconds, it amounts to some 18 cycles wasted each 8-hour shift.
18 cycles a day sum up to roughly 4,000 cycles a year. With a batch size of 50 kilograms 4,000 cycles equal 200 metric tonnes a year.
From here you can choose two ways to calculate the consequences, either work out:
• how much these 200 tonnes of “free” production could have been sold for (the mild one),
• or what the cost is of having the entire laundry wait for the batch washer to waste 4,000 cycles (the right one).
In both cases the sums are awesome – provided the batch washer is a bottleneck. But we never know that in advance.
Our machine designers and engineers have to assume that all machines are important to the laundry, every machine being a potential bottleneck.
So we should ask ourselves: When a batch is transported from one compartment to the next could any of the preparations taking place have been carried out before the change over… as e.g. draining or filling of water tanks, injection of chemicals, heating up or cooling down of the wash water, or the like? If the answer to this question is yes, by how much could processing time be reduced if processes were redesigned, and how much extra capacity would be released – the argument taken to its logical conclusion?
IED and OED
The principle of the calculation tells us to make a distinction between preparations, as well as handle at least one of them with great care and attention.
First of all we must demand from our suppliers and their designers that they discriminate between the two types of preparations (set-up): Inside Exchange of Die, IED, which either extends processing or requires the machine brought to a complete halt, and Outside Exchange of Die, OED, which can be carried out during processing without influence on start, speed or time.
IED and OED are not the best of names, but originating in the car industry 50 years ago they have become the common notions, so we still use them, also in other industries. Secondly, but equally important, we must demand from our suppliers that they convert as much IED as possible to OED. Finally we must be able to carry out the remaining IED as easy and rapidly as possible.
With increasing number of set-ups this distinction between IED and OED has increasing importance to a process route’s flow capacity. And there are many set-ups and change-overs in the laundries. Quite a lot actually.
Costs on bottlenecks
Calculating the cost of wasting 4,000 cycles taught us an important lesson about costs on bottlenecks:
The cost of an idling bottleneck (whether it be because it is being repaired, changed over, left unmanned during breaks, underloaded, processing products not demanded right now, products which could have been processed on other workstations, products already rejectable, or feeding a defecting downstream workstation) is the cost of leaving the entire laundry idle.
We must pay all possible respect in the laundry to these bottlenecks.
And while awaiting upstream workstations to send batches downstream we eat into the buffers. Like reservoirs they open the floodgates when the drought sets in.
But in reality there are only four economically motivated uses of buffers:
1. as planning points
2. to ensure the continuous processing on bottlenecks,
3. to minimize the number of micro pauses,
4. and to compensate for high set-up, change-over, idle or stop costs.
Long, closed process lines, without buffers between the workstations, force the planners to plan an unbroken flow of goods, from one end of the laundry production to the other.
If a workstation along such a process route breaks down, the flow of goods stops right out in the check-in, and all workstations along the process route stop consequently.
Buffers between the workstations give us the opportunity to shorten the batch flow, to dam up batches, and to re-route the batch flow down alternative process route sections, when needed.
In a limited economy the most important workstations to keep running are the bottlenecks. All the rest of the workstations, by definition, have higher capacity than the bottlenecks, and are able to catch up with them, once the batches flow again.
We know the cost of an idle bottleneck now, so buffers are used to make sure the bottlenecks keep on running, even when everything around them break down. Micro-pauses are short, unsolicited pauses imposed on workstations by interruptions in the flow of goods – sometimes caused by bad planning. Most often we don’t see them in the real laundry because the operators slow down their working rate to compensate for the interruption in the flow of goods, rather than move to another workstation.
Even the best product mixes are hard to plan in a buffer free production, though, without causing one or two micro-pauses, but with strategic buffers (and clever planning) we may be able to maintain an uninterrupted flow of goods. Finally high set-up, change-over, idle or stop costs force us to keep some workstations running, once they are started.
The cost of emptying a continuous batch washer is one example. To keep it running, once started, requires full upstream buffers. But buffers come at a cost. They drain our laundry’ lifeblood: The working capital. So here’s the point:
Unless the buffer contents form part of the laundry’s operating strategies and most of the categories have alternative process routes, buffers shouldn’t be filled at all.
In all other situations filled buffers are an indication of the planner being unable to take the entire production into consideration when planning, and make targeted plans accordingly. That’s how it is.
The cover up
Great batch quantities in the laundry, in front of every workstation, give back the degrees of planning freedom that specialization, constraints, dependent consumptions, and all the other bad things you may encounter in the laundry, have stolen from the planners. And yes, buffers and inventories are excellent means to get back degrees of freedom, but also only to the extent it is not possible by any other means to achieve the same – for instance by means of planning.
The laundry planners should only use buffers when they are able to explain and justify each batch in the buffers, and not just use them as means to cover up bad, insufficient, or downright lack of targeted planning.
Look at filled buffers as hazard flashers warning us of insufficient planning, capacity imbalances and financial drainage, and demand an explanation for each and every batch, in each and every buffer.
Buffer free planning
Of course, a production without buffers may give most young and unexperienced planners a cold sweat because every little planning mistake cause immediate and obvious consequences in the production, and later on in the accounts. In such a production all the consequences of planning errors, capacity imbalances and constraints surface.
There is no doubt that extensive use of buffers relieves the planning responsibility, but in return we get more than we bargain for. It is obviously more comfortable, but certainly not better. Not only do buffers drain the company’s working capital, they also hide hard facts – and spoil our chances to improve the flow of goods and of capital. We, the laundry managers, had better acknowledge that errors occur and have causes, and that it is our responsibility to try to eliminate them. Not drown them.
It is better to know of the problems and try to solve them, than to hide and forget them, and tell ourselves and everyone else that everything is just perfect. It’s not always true. The best planning result is achieved with a limited and motivated use of buffers, based on a buffer free production plan.
Degrees of freedom
One of the purposes of buffers is to give back to the planners the degrees of freedom stolen by machine design and plant layout. When we decide on the laundry layout, process routes, internal transportation systems and buffers, it is an important effort to try to keep down inventories and operating costs by preserving these degrees of freedom. We must avoid imposing restrictions on the flow of goods – like creating precendence constraints in conveyor systems.
Apportioned production
But not all laundries have the same opportunity to fill buffers. Some wash apportioned. They produce the linen and textiles collected from the customers and deliver the exact same items back again. The textiles circulate in a closed loop. This kind of laundry production is called apportioned production, because the laundry keeps each customer’s linen and textiles separated (apportioned) from other customers’ through the entire process. In other industries this production form is called order production.
Pool production
Other laundries break the loop, either by letting the customer herself or the drivers sum up the customer’s requirements, or by counting the collected articles in the check-in. Not the exact same items, but matching qualities and quantities are then picked in the dispatch department and delivered to the customer.
This kind of production is called pool production, because the laundry gathers all category identical items in big batches (pools), no matter from which customer they have been collected. In other industries this production form is called stock production.
Differences between apportioned and pool production
From exclusively producing in portions the laundry industry has developed into mainly producing in pools. And it is a natural conclusion to break the loop of circulating items. And break it precisely in the production because it allows the laundry to plan its day more or less independent of distribution, instead of being dictated by times, sequences and quantities determined by the market.
So the most obvious advantage of pool production is the possibility of producing in full loads. The check-in simply only send batches into the production when they are full. This is only possible, though, if the laundry has sufficient inventories to deliver from, while batches are being filled in the check-in.
Conversely, if a laundry produces apportioned, it will have to ask the customers to wait while their goods are being produced. The laundry stores the customers, so to speak.
The disadvantage of pool production
A major disadvantage of pool production is the disproportionately large inventories, the oceans of linen, we find in pool laundries all over the world. They are everywhere. With such quantities we loose touch, their sense of value, not only our own people in the production, but also our customers.
Only a few laundries know exactly how much linen they have in circulation, just as only a few customers know how much they have in stock. Billing systems are based on how much is delivered to the customer, disregarding how much is returned.
An article in Laundry & Cleaning News, May 1998, by Richard Merli, focused on the problem with these words:
“One of the most serious sources of financial draining for commercial laundries and linen rental companies is the loss of linen and textiles, estimated to be running at 1,000,000,000 US$ per year, in the US alone”.
Neither pool production, precedence constraints nor requisition systems should be blamed for these loses.
It is the entire industrialization process, with its specialisation, soaring volumes, increasing speeds and growing complexity, which has complicated the material flow beyond control.
But the amount (which only represent the loss of linen and textiles, not the other losses caused by bad planning and control) quantifies the importance of material flow control to a laundry’s finances.
The cost of inventories
Is pool production necessary at all? you may ask. No it is not.
All laundries have to deliver back to the customers what they have collected, in pool productions as well as in apportioned – unless of course the customer’s demands a change in productmix or volumes, as they do in and out of seasons.
But since the entire market most often experience the same seasons, there is actually no point in taking items out of circulation.
Inventory control focus on season peaks. The pool laundries only use the pools to reduce constraints and complexity and to ease the pressure on the planners. But like all kinds of inventories pools come at a cost, quite literally. And only if you are able to quantify these costs do you know the potential hidden in the pool laundry.
We know from experience that inventory management and material flow control in some cases have reduced a laundry’s textile procurement by up to 90% in 2 consecutive years.
An example: A small guesthouse (40 beds) didn’t bother to count the stock of bed linen, but ordered 500 sets at a time, every time the stock seemed low. Convenient perhaps, but expensive to the laundry.
In the chart you see why. The Y-axis is the number of sets and the X-axis is the time. In the first instance 500 sets are delivered to the customer, who use them over a long time.
In the other instance some 60 sets are delivered to the customer, who use them over a much shorter time. The darker area represents the sets sitting unused in the customer’s stock, i.e. locked-up capital and lost turnover.
Lot size and optimality
Which sends us back to planning and material flow control. The laundry planners should know of as many batches, categories, customer relations and batch preferences and for as long into the future as possible. If they know the net demand for an entire production shift, they have a good shot at making good, optimal plans.
I know. It sounds paradoxical to say good, optimal plans, because it implies “bad optimal plans” – but this is actually the case. With the right tools and methods you may be able to find an optimal plan, but imagine making a plan for just one batch.
Well, if you ask for an optimal plan for one batch, that’s what you get – with process route choices and allocation of the necessary operators to each workstation. Under the given circumstances the plan would be optimal.
But we cannot optimize plan interfaces, i.e. the borders between plans, because we do not yet know the content of the coming plan, and with only one batch in the plan, we do not have any alternatives connecting to the previous plan. We end up with nothing but plan interfaces. See?
More batches in a plan makes it easier to connect to the previous plan, and reduce the marginal weight of the interface to the subsequent plan. The more batches in a plan, the better we are able to avoid empty compartments, bath exchanges, jams, idle workstations, queues, idle operators etc. Time is important here too.
In the other end of the scale, the knowledge of all batches in a shift gives the planners the opportunity to assess and position each batch in the process sequence where it best fits in.
But most often we do not have the opportunity to wait until all batches from a shift have been sorted and categorized. We must choose our planning lot sizes somewhere between all and one batch.
In the next blog entry I’ll show how to search for the optimal plan, I’ll go through the concept of allocation effeciencies and I’ll discuss organization principles.
Hans arbejde er baseret på års erfaring med virksomhedsdrift og ‑planlægning på mange niveauer, både i produktions-, service- og udviklingsvirksomheder (bl.a. 8 års erfaring som vaskeridirektør).