Level Scheduling

seesawLean has it roots in manufacturing. It made the jump to application in health and social care years ago, but there is still much to learn from its use in industrial settings. This week, I had the good fortune to attend a visit to Nissan UK’s Sunderland plant, organised by Iain Smith of the North East Transformation System (NETS) and facilitated by Diane Miller of the Virginia Mason Institute.

The plant looks very much as you would expect a leading car manufacturer’s facility to look: it’s enormous, noisy, and filled with robots and people. The statistics are impressive: the lead time for a car is about fifteen hours from sheet metal roll to finished vehicle, and while the plant is open, one car is completed every minute. There are several aspects of the visit I plan to write about, but I want to discuss scheduling in this post. I’ll explain how I understand it to work at Nissan, and then consider lessons for health care.

In years past, car plants turned out cars that sat in great long rows in enormous car parks, waiting for someone to buy them: things are very different today. With Nissan, and I assume most manufacturers, every car that is build already has a buyer. This statement needs a pause for reflection, and deserves  repetition: every car they build, already has a buyer. This means that the purchase of the car triggers the car to be made at the plant. No building on spec, no rusting rows of cars, only cars that someone already wants, and for which they have agreed to pay.

The advantages of this for the manufacturer are obvious, but it is worth considering how this arrangement benefits the buyer. Assuming you are in the market for a new car, if you go to a Nissan dealer you specify the car you want – model, colour, finish, additional items etc. That order goes from the dealer to the Nissan scheduling system, and is eventually allocated to the plant to make.  They were behind when I visited, and the average time from order to delivery, I was told, was around six weeks.

The cars at Sunderland are built on mixed production lines – two models on one line, and two others on the second line, plus the new Infiniti, which was still covered in a not-terribly-convincing camouflage at the time of my visit. This again needs a pause for reflection: they don’t make one type of car for a week, on a line, then change the tooling overnight or over a weekend, then start making the second type of car. They make the same cars next to each other, on the same production line. So, for example (and Nissan will excuse me if I mix up the production lines – cars all look much the same to me) they might make a right hand drive Nissan Juke followed by a left hand drive Nissan Leaf, followed by a left hand drive Nissan Juke, and so on. All the specifications might change  – different engine size, different specification, one with a satnav, one without, and so on, so that there are thousands of possible variations of precisely what is built.

When you see it at close quarters, the pace of this is relentless. The staff don’t have time to think about the change from one car model to another: they have, on average, one minute to complete their task with each car. So you change from one car type to another, with no break – you clearly need to know exactly what you are doing.

As a job, it often looked difficult and demanding. Staff are multi-trained, and rotate jobs up to three times in a shift, which must at least help to reduce monotony. This multi-trained aspect is important in this type of system, and does read across to health and social care – more on that in a later post.

For the scheduling, they are using levelling (also known as heijunka) where rather than producing batches, the schedulers mix the cars in in the precise proportions that they have been ordered, so that the lead time from order to delivery, and the time for Nissan to bank the payment, is kept as short as possible. If they made one type of car for a fortnight, then switched systems and built the second type for the next two weeks, and so on, they would build in an automatic delay. By levelling, they level out the work for the staff, and also reduce the average lead time for their customer. I assume considerations like the timing of ships leaving to take cars to different countries also affects the scheduling, so it cannot be an easy job.

How does this all come together? The various car components are ‘pulled’ by the car manufacturing process. So, for example, engines were made on a parallel line and then appeared at the right time, to be fitted in to the correct car body. There were seven suppliers on-site, who made components and, like Nissan, they did not appear to build on spec, but rather in response to specific orders. So, for example, one manufacturer supplied the seats. They loaded the seats on to a moving line in the order in which they were to be fitted to the cars, and the system delivered them to the person fitting the seats in to the car, at precisely the time the relevant car rolled past their work site.

Although they could work with less warning, the Nissan staff explained that the current backlog of order meant that, about a month before the expected date of manufacture, the plant and suppliers can be told the rough expected schedule. A week before, they can be given a firm schedule that let them plan their work. This did not mean that parts were stockpiled: the body shop held car body parts that would run out within two shifts – less than a day.

What is the relevance of this to health care?

I took two main ideas from it. The plant did not build what they thought people might want: they built precisely what they had ordered. This was mass production, but it was personalised mass production. They built only in response to demand, and they built what was needed, and only what was needed. Stock, and therefore space, was kept to a minimum, and so reduced costs.

Clinics and operating theatres work in similar ways to production lines – some authors talk about ‘service lines’ rather than ‘production lines’, but the principles are similar. People with roughly the same type of condition – ophthalmological,  gynaecological, gastrointestinal – attend them, and we don’t tend to mix specialties within a clinic because the skills needed to help the person are often different.

Within specialties, the precise process a person needs will vary, but the general components of care are similar. We batch more in health care even within specialties – in ophthalmology for example, we’re likely to see all the people with macular degeneration at one clinic, the people with cataract at a second clinic, and so on – but in my experience we tend to keep the number of slots in a clinic pretty much the same from week to week rather than varying the mix as they can do in the car scenario, in response to demand. Scheduling it would be complicated, but if we were able to flex health care requirements even a little like this, so that we increased and decreased slots in response to demand, week to week, we would be likely to meet patient demands more quickly.

Secondly, the right item, right place, right time requirement was done impressively well. Again the health and social care context is different, particularly in emergency situations, but the general principle – of having the things people need to do their job available where and when they need it to stop delays and inconveniences to customers and to reduce frustration for staff – is obviously relevant.

The lead time for some of the items was very short – I was told that the final notification to the seat suppliers was around ninety minutes, for example. The ‘needs’ of the car, or at least of the car manufacturing process, ‘pulled’ the required items to it, so that no one waited if at all possible. This made me think about how we pull support to patients – if we know someone is going to theatre for procedure x, then we also know that they will need a recovery bay slot in, say, ninety minutes, a bed within two hours, a physiotherapist to help them up and about within 18 hours, discharge medication on average within three days, and so on.

Like the car example, we may move from estimated times to firmed up times with some changes, but the general pattern is known, and we can then adapt to the precise requirements. In some cases, we do seem to react as if a discharge is a surprise, rather than something we have known about for days. Thinking about the patient’s needs pulling the service they need to the right place at the right time, and going from a general expectation – probably ten people needing service y on Tuesday – to  precise final requirements – Mrs A at 9.50am today, Ms B at 10.35am and so on – feels possible, and may make the work of staff easier, as well as fulfilling the primary purpose of health and social care, of delivering value to patients.

Illustration courtesy of Idea Go at freedigitalphotos.net

4 thoughts on “Level Scheduling

  1. Pingback: What’s a Drum Buffer Rope System? | Lean Health Services

  2. Pingback: Visual Signals | Lean Health Services

  3. Pingback: Bypass and Baton Pass Zones | Lean Health Services

  4. Pingback: Book Review:Learning to See, by Mike Rother and John Shook |

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s