Degrees of freedom
We want good, optimal plans in the laundry production – how paradoxical that may sound.
A plan may be optimal, as well as every plan in a series of plans. It just doesn’t make the plans all together optimal.
Because of the continuous pressure of batches in the check-in, hygienic conditions and insufficient storing capacity the planners are seldom allowed the luxury to store batches from an entire shift, i.e. produce today what was sorted out yesterday.
More often the planners have to decide on a planning lot size between one batch and all batches. This decision is normally determined by balancing the market situation, the physical storage capacity and the finished goods inventories.
Furthermore it is not suitable to plan ahead too long. The world changes. A machine breaks down, operators are taken ill, new rush orders arrive, and so on. Contrary to short production plans, long plans involve the risk of something changing during execution, which calls for replanning. Short plans do not have to be replanned. They end in a moment anyway.
But if there is a major imbalance in capacities, for instance an oversized continuous batch washer (CBW), when do you start the CBW, and when do you stop it again? How big should the production lot size through the CBW be, when the machine is so expensive to stop, but inexpensive to run once started?
Had we had the opportunity, had we had the time, the computing power and the software, we could have taken all the batches, all the capacities and all the employees from a given day, and calculated the consequences of each and every sequence alternative – with all the constraints and dependent consumptions our laundry production is subjected to. It is from these process route alternatives and batch sequences the optimal production lot size on a given workstation is derived.
In a dynamic environment that is, what it takes to calculate the right lot sizes. There are no easy formulas. Those that exist don’t work, are oversimplified, or give, at best, a suboptimal solution to a limited part of the problem.
But we do not have the time or the computing power. To that the problem is far too complex, far too big. Instead we have to search for the optimal solutions in other ways.
And searching is the right word for it, because the optimal solution is constructed. It is not an outcome, which can be calculated or localized, or a decision pattern, which can be reproduced.
It is created. In the dynamic production environment it breaks out as the result of coordinated, purposive, focused actions, decision by decision, and is identified by its characteristics.
The optimal solution is created by selection, and spreads its consequences to consumptions and capacity loads, and from there to the accounts.
We know from other batch industries that the total variable costs depend on the very way we send batches through the production – because every batch draw differently on the resources, because alternative process routes have different flow capacities, and because batch sequences in this way indirectly determine the resource consumption and capacity utilization.
Once and for all we have to understand what the optimal solution looks like, and what its qualities are, in order to recognize it by its characteristics when it surfaces in the laundry production. And in order to choose batch sequences, process routes and employee allocations in such a way that we, by each choice, make sure that it really is the optimal solution we create in the production, when our decisions are carried out.
When 10 different batches may be sent into the laundry production in more than 3.6 millions different sequences, that is, what it is all about.
The unpleasant questions
Not all these sequences generate unique consequences, but most of them do. The problem is that those responsible for the production planning most often are not aware of it, and are almost never able to point out the sequence which best fulfils the management’s requirements – whether it be regarding economy, quality, lead-time, consumption or load.
And then it gets unpleasant.
Did the laundry management at all give the planners the necessary information to make qualified decisions?
Do the planners know what is required of them?
And of the laundry? Do they have a proper education?
The necessary knowledge of the production and its dependencies? Are they able to foresee consequences of their decisions?
In time? Are they rewarded for good decisions and reproved for bad ones?
What is a good decision?
What is a bad decision, and how do we recognize it if it should appear in our production?
To give specific answers we need a specific laundry – but generally only a few are able to respond with a definite yes to these questions.
When asked most responsible planners and production managers answer no:
We do not have access to the relevant information, methods or systems to carry out production planning according to specific targets,
We have not been informed about the connections between the daily planning decisions and the consequences in the accounts, or their weights.
The people responsible for the total variable costs in the laundries worldwide (some 10 billions UK£) have – at best – a craftsman’s background and 3 months general laundry education, with no qualification in planning techniques or insight into financial contexts,
It is most often unwelcome and almost reprehensible to reward and reprove in the production. In most laundries, especially in Western Europe, they avoid it. It is more or less analogous to ban counting the goals in a soccer match, or to clock a 100-metre race – events most of us take lightly. It is all right to have fun in the production, but make no mistakes. We do not run the laundry for the fun of it,
We cannot tell a perfect plan from a good plan, and sometimes not even a good plan from a bad. We do not know what to look for. Of course, afterwards everybody can tell a good day from a bad, measured by productivity or consumption, but not the good decisions from the bad ones while they are made or carried out,
Most often there are no systematized or formalized planning strategies in the laundries, and if there are, they most often do not cohere with measuring points, measuring frequencies, key figures or the desired results in the accounts.
The problem is, that when you look at a plan isolated, nothing tells us whether we are actually dealing with the optimal plan or not.
We have to know the cost curves if we are to identify the solution in the bottom of the curve – the one with the minimal total variable costs, shortest lead-time or lowest water consumption etc.
But do we know the curves?
No. They are unique to each planning situation. And we do not have a figure or a value to tell us unambiguously that a plan is optimal, i.e. cannot be improved. Only experience, insight into planning techniques or a set of special key performance indicators (the allocation efficiencies), which we are not able to compute manually.
Instead we can try to re-establish one of the really good days as a plan in a Gantt-chart. It is an exercise that gives a good general view of the laundry production’s dependencies and preconditions, and allow us to recognize patterns which we can use as product mix benchmarks in future plannings.
Plenty of time and access to all the batches from a production shift would allow us to sort, sequence and route batches as they best fit the available capacities, bottlenecks and buffers, and allocate operators accordingly.
But even in such an advantageous situation we are not able to take into consideration all the constraints, dependencies and alternatives. There are too many. We would also require sufficient raw computing power and an optimizer software called APS, Advanced Planning & Scheduling System, which costs a fortune.
For lack of an APS we could use operation strategies based on the same search techniques used in the APSs. The operation strategies are simpler and do almost the same, but they are hard to handle manually and require educated and experienced planners and managers. And the strategies themselves… well, not that many people in the world know how to formulate them. But still, education is often cheaper than management software.
For lack of an operation strategy we could use product mix norms as an approximation. They identify efficient category patterns and batch ratios down the laundry’s process routes; constitute a kind of benchmark for choosing, sequencing and routing batches into the laundry. But, but, but… they are static solutions to dynamic problems. We may be able to identify such patterns, but if something in the laundry set-up changes: consumptions, capacities, availabilities, demand etc., they are no longer valid. And things do change, needless to say – sometimes by the hour.
For lack of a product mix norm and any other focused method, we try to make each batch fit in as best we can. Maybe two, three or four batches at a time, but to do so, and still be able to avoid micro and macro pauses, we need a lot of buffers. Many laundries are operated in this way.
And there is absolutely no doubt: In a dynamic market and a planning environment subjected to all the variables in a modern, complex laundry production, optimization with an APS is by far the easiest, fastest and most efficient means, and actually also the planning method which require the least experience and technical background, because search strategies and planning techniques are built into the systems.
On the other hand operation strategies and product mix norms are applicable almost immediately, without the big investments in computers, network and software. They are nothing but condensed knowledge.
And they do exist, the few really skilled laundries, those who use operation strategies – those who have chosen a specific criterion (e.g. to minimize costs or lead-time, or to maximize productivity) and have formulated a strategy, which they have reason to believe will fulfil this criterion.
But what is an operation strategy then?
It is a formalized method for how to:
• define an overall, specific criterion for a production plan,
• derive from this criterion a search strategy that creates the optimal arrangement of activities in the laundry production, and
• define key performance indicators to ensure, without delay, that the optimal solution actually is materializing, batch-by-batch.
Operation strategies are based on the recognition of the optimal solution’s characteristics, much like the search strategies found in APSs, just not as elaborate, though.
Once the planners have realized what the characteristics of an optimal plan are (through real life experience, practices with Gantt-charts, from a simulation model or from an APS), they also know what they want to see in the production. With knowledge of planning techniques they should be able to formulate decision strategies that re-creates these characteristics.
Without going into full length or detail, a concrete operation strategy aimed at minimizing lead-time could read as follows:
Among all the sorted out, categorized and prioritized batches in the check-in:
• chose an adequate number of batches (e.g. 20 or 30)
• whose weighted, average drying time
• divided by their weighted, average CBW cycle time
• are close to, but do not exceed, the available drying capacity.
It may sound difficult, but it is quite simple once you get the drift of it. And it prevents dryer jams.
From this pool of batches chose the one:
1. whose specific drying time / cycle time ratio best level the average ratio at the available
drying capacity (avoids dryer jams),
2. if two or more batches do that equally well, chose the one(s) with the least influence on the
resulting CBW cycle time, (groups batches around equal cycle times)
3. among these, chose the ones which have the earliest end time on the last workstation along
process routes (favors fast batches, low load process routes, parallels the flow of batches,
takes the pressure of bottlenecks and keeps a high average lead-time),
4. among these, chose the one with the earliest start time on first workstation downstream from
the dryers (favors process routes with empty or low load buffers, and parallels the batch
5. and repeat from 1. until all batches are arranged in sequence.
The main purposes of this strategy are to parallel the batch flow, aim it at the fastest process routes and send batches downstream as soon as possible to achieve full overlap. This is how lead-time is reduced.
Depending on the planner’s skills and the information available, the strategies may be more or less detailed. Lowering the level of detail would be to focus on buffer contents only. Raising the level of detail would be to include rush order priorities, keep together batches from one and the same customer, split process routes up into sections betweens buffers, and the like.
In this strategy I have assumed the planners know what to sort in the check-in and what is required in the dispatch, i.e. they have determined the net demand.
And as you may have noticed, consumptions and costs are not included in the strategy.
Had the strategy criterion been different, e.g., minimizing the total variable costs, the strategy focus would also have been different. We should then identify batch sequences with low dependent consumptions, workstations with low set-up and stop costs, process routes with the few operators, and so on. We would see sequences, route choices and allocations different from the ones generated by the first strategy.
Each criterion calls for a unique strategy. And since the laundry, some days, prefer to minimize costs, where as they at other days want to minimize lead-time, they need to formulate strategies to suit the alternating requirements.
Among the blasted
We all know them, the days when everything goes well together. The right linen is at the right place, at the right time. There is a calm, steady flow of batches through the production, in pace with dispatch.
Everyone having worked in a production knows that among all the other blasted, endless and stressing days, these quiet but very productive days occur.
In a Gantt-chart, one of these better days might look like this:
Each beam represents a batch, and the colour of the beam represents the workstation.
It’s a stable plan, with efficient operator allocations, no nervous reshuffle, and only a few short micro pauses – even though the plan is buffer free. It couldn’t be much better.
In a laundry with no buffers, and where the operators keep a high, steady pace, you will find the small pauses that occur, when an operator awaits batches on a workstation for a few moments, or a when a workstation awaits operators, or when a batch awaits processing.
Employee, workstation or batch pauses.
From the micro pauses we are able to calculate a set of very powerful key performance indicators (KPIs), and even though they are hard to quantify, understanding them gives us a fundamental knowledge of what is required to make optimal plans.
One laundry produces 35 kilograms per employee working hour, another 53. Which laundry is getting the most out of its potential?
We don’t know. It depends on how much potential is left – wasted. Productivities do not tell us that.
To find out we need to deduce a method to quantify the locked-up potential, like weighing the waste steel plate from plasma cutters.
In a laundry waste is measured in minutes, i.e. the time batches, employees and workstations would be inoperative had they been planned in a buffer free production.
EAE, RAE and BAE
The arithmetic exercise gives us 3 key figures:
• Employee Allocation Efficiency (EAE), which expresses to what extent one or more employees have been operative during the time they have been available to production,
• Resource Allocation Efficiency (RAE), which expresses to what extent one or more workstations have been operative during the time they have been available to production, and
• Batch Allocation Efficiency (BAE), which expresses to what extent one or more batches have been processed during the time they have been available to production.
Together they are called Key Performance Indicators.
So here is my point: Optimality requires high values on all three KPIs – at the same time. Not the easiest thing in the world.
To keep a high EAE-value, we must avoid reallocations, waits and jams, and maintain stable buffer contents.
To keep a high RAE-value, we must reduce the number of set-ups, starts, change-overs and jams, and keep stable buffer contents.
To keep a high BAE-value, we must maintain a steady, parallel, overlapping batch flow, avoid jams, and keep buffers empty.
Productivity calculations do not tell us whether we are using the laundry to its full potential, or how to do so.
The Key Performance Indicators do that. And they tell us where to focus our attention and efforts, be it on the employee allocation, the batch allocation or the resource allocation.
And here’s another point: We will only find out, if we start planning buffer free. Good plans with buffers are based on good plans without buffers.