ERP capacity planning and scheduling
Discussions on capacity planning and capacity scheduling can be difficult because some people confuse the two and some believe the terms to be interchangeable. However, they are two different things that solve two different problems. Capacity planning exists to help ensure that capacity is available when it is required, and capacity scheduling exists to help optimize whatever capacity actually is available. The former is primarily a medium to long term tool whilst the latter frequently only looks a few days ahead; perhaps a couple of weeks at most. A look at both to see what they can and cannot do will be useful in helping to decide if individual companies need planning, scheduling or, indeed, both.
Sometimes called Infinite Capacity Planning or Rough Cut Capacity Planning (RCCP for short) is often the easier of the two to implement and to run.
As mentioned above, this is a medium to long-term tool to help ensure that future production plans are viable and to give advance notice of when additional capacity needs to be turned on; either permanently through the introduction of additional plant, or temporarily through the addition of overtime or extra shifts. It is not the job of Capacity Planning to sequence or optimize jobs in production (that is the task of Capacity Scheduling) and so finite accuracy can be traded for speed; particularly as, when looking at future requirements, it will generally be looking at forecasts and these by their very nature rarely approach 100% accuracy. Its job is to report in weekly time buckets to identify if and when extra resources are required.
The base data, on the other hand, does have to get down to hours and minutes so that the results are, if not precise, at least accurate. To do that, it has to multiply the amount of work to be done by the amount of time that is required to do that work. Let's say that the forecast calls for 1000 Type One widgets to be made in the first week in January. The routing might say that to make a widget, one hour is needed in the machine shop, one quarter of an hour in the paint shop and a quarter of an hour in the packing area. Multiplying that out gives a requirement for:
- 1000 hours in the machine shop,
- 250 hours in the paint shop, and
- 250 hours in packing
To make Type One widgets in the first week of January. Note that it doesn't matter which day, or days, in that week that the widgets will be made on. Apart from anything else, the plan is probably being run so far out into the future that any such accuracy would be illusory. Some RCCP systems, though, do recognize that operations at the beginning of the manufacturing process will, at least in the case of long production lead-time items, take place in earlier weeks than final assembly and packing, and these may produce better results for some companies.
RCCP will then calculate the time requirements for all of the other things that are planned to be made and the result will be a report that shows whether or not the plan has generated a realistic demand on factory resources. Frequently, capacity utilization will also be displayed as a percentage of availability so that resources that are critically loaded can be quickly identified.
If there is an insufficient manufacturing resource to meet the plan, there are only two alternatives:
- Increasing the resource available, by adding extra shifts or plant, or
- Changing the plan.
For RCCP to work it needs an accurate picture of what is to be produced and an accurate picture of how long it takes to produce those items. This latter requirement may not be as easy to satisfy as it sounds, for three reasons.
Firstly, many companies, with the death of time and motion study, do not have accurate information on how long it takes to make their products. There may have been timings created for costing or for bonus purposes but are they really accurate? The first task, when introducing RCCP is, therefore, to get good, reliable timings. If that can't be done, it probably isn't worth continuing. You best that can be achieved, using flawed or inaccurate timings, is to ascertain that a department or machine is producing, say, 1200 hours' worth of output per week (regardless of the actual hours worked) and use that as the capacity norm, but such a crude approach will fail if the timing inaccuracies are not uniform across the product range.
The second problem to be overcome is to know how many hours per week usable capacity actually are available. A particular machine might nominally run 8 hours a day, five days a week but there will be breakdowns, interruptions and all manner of unplanned events. When assessing capacity availability companies need to compare against 'demonstrated capacity'. Analysis of past performance will show how many hours a day a machine or person actively produces. Frequently, in real life, this is about 85%: that is 34 usable (demonstrated) hours out of a theoretical capacity of 40.
Lastly, in many industries there are still significant changeover times between jobs. These changeover times may not be fixed - it is not unusual in, for example, the plastics and food industries for changeover times to be product dependent. It takes longer to move from chocolate ice-cream to vanilla than it takes from vanilla to strawberry and it may not always be possible to model these changes far out unless there is a set cycle of changeovers.
What makes capacity planning easier to set up and operate than capacity scheduling is that, in most factories, it is unusual for there to be capacity problems in all work centers or on all machines. Frequently the 80/20 rule applies: that most problems are caused by a small number of resources. Planning staff probably recognize these already as being bottlenecks so can focus their attention on these areas, particularly when it comes to calculating or measuring accurate process or run times. This will get the system up and running faster and the reality is that companies generally don't have to worry about workstations that are significantly under-loaded.
Before leaving the topic of capacity planning; one last thought: capacity is usually considered in terms of time but there can be other constraints. Space (whether for the assembly of large items or for the storage of stock built ahead of sales requirements) is sometimes an issue. Likewise, and particularly but not solely in the case of stock-build, working capital can be a constraint. Companies will be procuring materials and maybe paying for them some time before they get paid by their customers for the finished items so would be wise to ensure that a profitable business does not sink in the swamp of cash flow.
This is sometimes called Finite Capacity Planning or Finite Capacity Scheduling.
Whilst Capacity Planning (RCCP) is a medium to long term tool, the short term is the prerogative of Capacity Scheduling. At this point, it is for most companies too late to solve capacity problems by installing extra plant or by recruiting extra staff (although short term overtime remains a possibility, albeit it one that can incur an appreciable cost overhead). What is necessary now is to optimize the use of the capacity that they have.
To do this, they again need accurate detail of the tasks to be performed, the resource (people and/or plant) that will perform those tasks and the amount of time that is required to perform the tasks. Where capacity scheduling becomes more complex is when capacity planning is in the options that now come into play. RCCP, being in time buckets of up to a week, would be happy to say that there was sufficient capacity in a five day week to both assemble and pack an item, but what if the assembly could not be completed until right at the end of the week? Packing capacity that was available earlier in the week (and satisfied RCCP) cannot be used (disappointingly, time only travels in one direction!). So capacity scheduling needs to consider, not only the time that operations take, but also their sequence.
Again; this may not be as easy as it sounds. Obviously items can’t be packed until after they have been assembled but, in some manufacturing processes, operations can take place in parallel or in varying sequences. For example, if manufacturing wooden ladders, it might be normal to wait until the ladder is assembled before varnishing it but the individual pieces could conceivably be varnished before assembly if they had to be. So in addition to ensuring that the data that capacity scheduling needs is accurate, it is necessary to ensure that the ‘rules’ that are fed into it are not only accurate but also complete.
There are other things to consider. Does one operation have to be completely finished before the next can begin or can operations overlap? For example, if assembling and packing one thousand widgets, is it necessary to wait for all one thousand to be assembled before packing can commence, or can packing begin after the first, say, one hundred have been completed? The rules will likely change from item to item and will have to consider factors like comparative run speeds of the two (or more) processes: i.e. packing could theoretically commence after one hundred widgets have been assembled but, because if packing is faster than assembly, that department would soon ‘catch up’ and run out of work.
There is also, in many industries, the very significant factor of set-up times to consider. Although many companies for many years have been working hard to reduce machine set-up times, some things cannot be avoided. Going back to the ice cream analogy, changing from vanilla to strawberry can be done very quickly just by adding the strawberry concentrate and flushing though the first few liters as waste but, when changing from chocolate flavor back to vanilla, companies have to do a complete clean-down of the machine. Similar problems occur in the plastics industry and doubtless others.
The system must be set-up to ‘understand’ these factors; knowing that a perfect plan may be impossible. Minimizing set-up times will increase throughput but it may do it at the expense of making some jobs late. Where is the balance to be? Obviously, if there is sufficient capacity for all requirements to be met on-time and with minimum set-up then there isn't a problem but, following a machine breakdown, or a batch of faulty materials from a supplier causing a higher than normal reject rate, there may be two or more jobs competing for priority at a particular resource.
Some companies respond by saying, “The biggest customer gets priority.” But what should happen if their three biggest customers are worth £7m, £6m, and £5m respectively to them and the biggest customer had already been given priority twice already this month at the expense of customers 2 and 3, who were now getting to be unhappy? What if giving the biggest customer (worth £7m) priority again was to cause the other two to walk? Individually they are worth less to the company than customer 1 but, collectively, they may be about to lose £11m worth of business to keep £7m. Do the rules now need to change?
In summary then; what is needed for a capacity scheduling system to work are:
- Accurate data in the Routings file
- A knowledge of when and how different operations can use alternative machines (e.g. can a job that is generally run on a CNC machine be transferred to a less-efficient lathe if a shortage of capacity so dictates?)
- A knowledge of how set-up times can change according to job sequence (e.g. vanilla to strawberry ice cream and chocolate to vanilla)
- A knowledge of which operations can overlap, and by how much
- Accurate data about machine capacities.
In conclusion; capacity planning and capacity scheduling are two different things; although capacity scheduling systems have come down in price drastically in recent years and, if implemented intelligently, are frequently fast enough to perform the planning role also. But capacity scheduling is more difficult to set up and run because it requires much more data than capacity planning and it needs that data to be very accurate. So the choice for some companies will not be easy.
Featured white papers
ERP Software Pricing Guide
Get the latest pricing information on over 80 popular ERP systems, and learn how to budget for your ERP project in our free guideDownload
60-Step ERP Selection Checklist
Get the comprehensive checklist for your ERP selection projectDownload
ERP Implementation: 9 steps to success
The 9 proven steps you should follow when implementing ERPDownload
Why traditional scheduling systems don't work for custom manufacturers
A guest blog from Genius Solutions discussing ERP and custom manufacturing
Why a food specific ERP system is a must-have
Key features and requirements food companies should consider when searching for an ERP
5 Ways Big Data is Changing ERP Software
Discover the ways in which Big Data technology is changing how we use our ERP systems from CRM to...