Start-up & shutdown production simulation

At Production Support 56, we specialise in process improvement and manufacturing simulation. We have modelled many production lines and generally find that when a line is up to speed that it is relatively efficient. The more fruitful opportunities to improve are found outside of normal operations, when things go wrong or when line needs changing. An often-neglected area for improvement is the start-up and shutdown of a production line. This is the period at the start of a day, or campaign, where the machines are turned on and materials starts to fill the line and then the reverse for shutdowns. As the line cannot run at full speed during these periods it impossible for the line to be as efficient as during normal operations. These inefficient periods can last minutes, hours or even days. Manufacturers often compensate for this by increasing batch sizes and campaign lengths which results in over-production. Simulation is a great tool to assess different production philosophies and help determine the trade-offs between batch size and efficiency, but today’s article is going to focus on start-up and shutdown of production lines.

Early model considerations for production simulation

As always, before starting any model you should define the question your model is attempting to answer. If you are only interested in the normal operation (when things are working as expected and the line is at normal speed), then don’t waste any time worrying about the start-up. I have discussed defining a good model question in a previous article. Most simulation software has a warmup period where the data generated during the start of the model is ignored. This is to prevent the start-up data from distorting normal operation data. This is correct as they are two distinct phases of a production line. If you are only interested in normal operations, you should also make sure that the data generated during shutdown of the model is also ignored. This is often easily achieved by having the model abruptly stop when it is in full flow. We use Simul8 simulation software for our modelling, it has a built-in warmup period function but depending on the application we generally program the model to collect data in distinct phases.

I often refer to normal operations as the steady state of the model, this is where the inputs and outputs of the model match and the line loadings and work being done is relatively constant. This definition helps define the transition from the start-up to steady state and through to the shutdown phase.

To complicate things, there are two types of start-up: cold and warm. A cold start-up is when the equipment is completely off and requires a series of activity to get them operational, for example taking a furnace from room temperature to operational temperature, there will be a series of complex and time-consuming activities to achieve this. A warm start-up is where equipment is in an idle state and require minor activities to get it operational, for example a spiral freezer may be at the correct temperature it just needs the conveyor starting. For your model you need to be aware of which type of start-up you are simulating to ensure you collect the right information.

If you think start-ups and shutdowns may have a significant impact on your model’s purpose, then it is worth spending the extra time broadening the scope of your project. This means you must accurately model the start and end of a production campaign and program the software to collect data for each distinct phase. The second thing to consider is which parameters are the most important and will affect the models results?

Relevant factors for production start-up and shutdowns

When planning data collection for a model or process improvement project I consider the five main factors which make up the well-known 4M’s and 1E, Man, Machine, Material, Method and Environment. It is important to consider each factor and identify the key levers that affect your model, and which are important in answering the question.

Production simulation levers

Environment: Manufacturing processes are normally conducted indoors with important environmental factors controlled, e.g., light, temperature, humidity, and air flow. However, it is worth considering, especially if the process you are modelling is outdoors, but for most modelling projects it can be safely ignored.

Labour (Man): The important data here are shift patterns, breaks times, manning levels and skills. Quite often there will be key people who manage the equipment and materials at the start and end of a production campaign.

Machines: Some machines require a warmup period (such as freezer or fryers) or require calibration or setting up. At the end of a campaign, equipment may need to be made safe or cleaned down. These are activities in their own right and may require specialist skills but will only occur once at either the start or end of a campaign.

Materials: Some materials can be left in a semi-complete state. These are generally discrete non-perishable parts (e.g., a car engine), at the end of a production shift the current activity will be complete and the part can wait on the line until the next day. Other materials may need special storage at the end of a day especially if the material is valuable or hazardous. Some materials will be perishable (such as foods) and so the line may be emptied and cleaned at the end of the production shift.

Method: The method will define the cycle times and skills needed to start-up and shutdown a line. The method (or campaign philosophy) may require a build-up of part finished materials (work in progress, WIP) during the start of a campaign and for that WIP to be run down at the end of the campaign, or for WIP to be stored at the end of a campaign and brough back out at the start of the next campaign. Tactical use of WIP can help speed up start-ups and ensure the smooth running of the line.

Real world case study with start-ups and shutdowns

We recently modelled a melt and cast process, where the cast-pieces where machined and the removed material was returned to the melt process. They ran in two week campaigns (service life of crucible) and it took about a day to get the equipment up to temperature and about eight hours to cool and clean the equipment at the end.

Seed material

The melt material was made up of 90% new and 10% seed material recovered from machining. To make life interesting the machined material was not generated until 60 hours into production and it was extremely toxic. So, for the initial melts, a stock of recovered material was required to seed each melt, once the cast parts were ready for machining then the production line generated sufficient recovered material to seed each subsequent melt, and a steady state was achieved. The last casting process took place ten days into the campaign, after which, significant amounts of recovered material was generated but not consumed. The graph shows the amount of toxic recovered material in storage and has a bathtub-like curve with three distinct phases of production.

Results from production simulation case study

The model allowed us to work with the process stakeholders to minimise that amount of toxic recovered material held, and ensure there was sufficient in stock to start the next campaign.

Start-up and shutdown

One of the interesting features of this operation was that there were a lot of activities required to start the production campaign. Modelling allowed an assessment of different start-up schedules. We found that there was sufficient free labour at the end of a campaign to prepare the furnace for the next campaign, this included emptying, cooling, stripping, rebuilding and setting. Furthermore, it was found that the machining equipment could be prepared at the start of the next campaign as the first melts and cast activities were being carried out. This approach taken was similar to single minute exchange of dies (SMED) methodology (I hope to discuss this in a later article).

This optimisation of the start-up and shutdown activities significantly reduced the need for overtime, minimised delays, and increased the capacity of the plant, all without spending any money.

Summary: Effective production simulation

We have found that manufacturing simulation is a great tool to support process improvement and operational design. The key to creating a good model is to base it on a single question which incorporate the goal and the levers. For example, ‘Can I increase the throughput of the line by optimising the shift pattern?’. When you are planning the model, you should consider if start-ups and shutdowns affect the goal, and if so, which are the key elements. If you are incorporating start-up and shutdowns, you should make sure the model collects the data separately for each distinct phase.

With a model that incorporates start-ups and shutdowns you can determine the full operational costs in terms labour, materials and energy. You can then run various scenarios to optimise the trade-off between warm start-ups and normal operations. This is especially relevant for energy intensive equipment like freezers, you may want to idle them between production campaigns at a higher temperature whilst not detrimentally increasing the start-up time.

You may find that bigger impacts are achieved outside of normal operations, and if you don’t model them, you will never know.

Related production simulation content

If you want to learn more about production simulation you might want to check out my previous publications on SCDA:

You May Also Like

Leave a Reply

1 comment

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.