This post is part of a 4-part series on Capacity Planning. Read the introduction and posts on Long-term Capacity Planning, and Mid-term Capacity Planning.
Short-term capacity is all about details. Now we pay attention to specific products being manufactured and no longer use aggregated data. This is when we pay much more attention to each part in the bill of materials (BOM) and each step in the routing operations to ensure we have materials, labor, and machines available on time. When we have one, the MRP tool (Manufacturing Resource Planning) can be very useful here, as it can suggest master schedules based on demands, BOM, and routing data. We have a simplified example of this kind of time-phased demands and supplies table below, for some dummy item where all parts are always in stock, so the only lead time is the manufacturing lead time:
Figure 1. Master schedule when manufacturing lead time=2 weeks
In the simplified table above, since the manufacturing lead time is two weeks, anything within the next two weeks as of today is already scheduled to be produced. Everything after two weeks is only planned to be produced – nothing released yet on the shop floor. We can see that we have an order expected to be released two weeks before each planned order receipt.
This is how far the MRP can go, as it is often not considering capacity limitations but only standard lead times. If we are in a chase production strategy, where capacity is highly flexible, and we want to avoid storing costs, this is probably fine. But a chase production strategy is not always realistic. It can be tough on workers’ morale and loyalty, or expensive in equipment. In some contexts, it might be best to level production and plan to store items in low-demand periods to use the resources as close as possible to full capacity to have more profitable investments.
When capacity limitations are considered, maybe the same table is limited to 2 weekly production batches (80 units):
Figure 2. Master schedule concerning capacity constraints
How can one turn the unlimited capacity table into the limited capacity one? With CRP (Capacity Requirements Planning) tools! A CRP would typically dig into each operation of the product routing to detail the required capacity at each work center. It would offer many views to show both the future orders to process and the work centers’ usage. The production planners can then use the CRP to generate what-if scenarios and tweak the MRP planned orders on a graphical user interface. Once they are satisfied with the schedule they have established for a given period, they can firm orders if they don’t want the MRP to reset them on its next run.
Let say the item above (#987) is fully manufactured at work center #34 in a single operation. However, other items (let’s all consider single operation orders for simplicity) might also be using the same work center, making it very busy, hence the weekly limit on 80 units of item #987:
Figure 3. Simplified Work Center Load Profile
As we can see, the work order #5 cannot be fully processed on Tuesday, on top of work orders #3 and #4, so the production planners could do either of the following:
- use the free shift on Monday night to start processing work order #3 earlier
- process work order #4 a day ahead of time
- split work order # 5 between Monday and Tuesday.
A graphical tool such as a CRP can facilitate these decisions. Even when you have a more complex scenario where different products go through many work centers, you can see the holes in the schedule and tweak them. And as you move orders around in the work centers, the CRP would update operations schedules so that the following operations can start earlier or later. You can also get warnings if an operation cannot be executed earlier because it has a predecessor that wouldn’t be completed yet.
What if planners don’t have many holes in a work center schedule, at least none early enough to split overloads and still meet demands? Time to prioritize! There are different rules to prioritize operations:
- Don’t prioritize
- First in, first out!
- Prioritize shortest operations
- Get as many jobs done as possible in the shortest time. It doesn’t necessarily mean that we’ll optimize the number of work orders completed on time, and we may build up work-in-progress inventory between work centers. Longer orders will likely see delays.
- Prioritize operations based on their work order’s due date
- Simple. However, it doesn’t give any insight into how much time we still have ahead of us to get the order out (how many operations are still do be done at other work centers after, and how long do they take?). Shorter orders might get delayed by longer ones.
- Prioritize operations based on their due date
- It sounds like an efficient use of the work center capacity. However, we lose sight of their related orders due date, and we don’t consider the operations processing time.
- Prioritize based on critical ratio
- Critical ratio is based on the time remaining to get the order out divided by the time required to get it out:
Critical Ratio=(Due Date-Today)/(Remaining Lead Time)
Operations belonging to the orders with the lowest critical ratio are then prioritized since critical ratio values have the following meanings:
- CR<0 ⇒ The order is already late
- 0<CR<1 ⇒ The order is behind schedule
- CR=1 ⇒ The order is right on time
- CR>1 ⇒ The order is in advance
With the critical ratio, regardless of how operations were moved around already, we never lose sight of how soon the related order must go out AND how much time we have left to do it.
- Slack time: time remaining to get the order out – time required to get it out
- Another way to look at similar information as the critical ratio!
Notice the load profile above is currently looking at a single week. CRP tools can display the load on different time ranges so that an overload on Monday of week five can be foreseen and addressed early. What is early enough? This is highly variable. It is usually recommended to plan manufacturing ahead as far as the longest cumulative lead time (lead times for parts + longest operations sequence) for a product. Unfortunately, orders may not be placed early enough by customers to plan soon enough, so sales forecasts could be used instead of actual orders. This can be illustrated with “water” zones:
Figure 4: Demands Water Zones
The frozen zone is the time range in which no new requests from customers can be accepted since production has been scheduled to meet actual customer orders, and all parts required have been procured and committed to the various work orders. A change in any product’s production schedule in the frozen zone would be disruptive, and upper management approval would be required to modify anything. The demand time fence that distinguishes the frozen zone from the slushy zone is usually determined by upper management.
Production planners can rearrange priorities in the slushy zone, however, these changes must be managed manually – the MRP should no longer modify planned supplies in this zone, as not just any change could be made! The planning time fence that distinguishes the slushy zone from the liquid zone is based on the longest cumulative lead time. Our work center load figure above would be in the slushy zone.
Finally, the liquid zone is where anything can happen. Production planners can let the MRP reset planned work orders as the demand continues to evolve in this far away future zone.
Make-to-order (MTO) – Example:
I’m no expert on toilet paper manufacturing, so please be indulgent when I start throwing random numbers in this example to illustrate the manufacturing concepts discussed.
Back to the toilet paper manufacturing example: with all the details jumping in, you’ll notice this is starting to be messy!
Now that the production planners have performed their mid-term capacity planning, they communicated to the sales team how much they can promise.
For the next 4 weeks:
Type | Capacity (kg/day) | Capacity (rolls/day) | Capacity (8 rolls packs) | Capacity (16 rolls packs) | Capacity (24 rolls packs) |
2-ply (85g) | 123,250 | 1,450,000 | 61,250 (41,650 kg) |
30,000 (40,800 kg) |
20,000 (40,800 kg) |
3-ply (125g) | 30,000 | 240,000 | 5,000 (5000 kg) |
5,000 (10,000 kg) |
5,000 (15,000 kg) |
For the following 4 weeks:
Type | Capacity (kg/day) | Capacity (rolls/day) | Capacity (8 rolls packs) | Capacity (16 rolls packs) | Capacity (24 rolls packs) |
2-ply (85g) | 170,000 | 2,000,000 | 90,000 (61,200 kg) |
44,000 (59,840 kg) |
24,000 (48,960 kg) |
3-ply (125g) | 30,000 | 240,000 | 5,000 (5000 kg) |
5,000 (10,000 kg) |
5,000 (15,000 kg) |
The sales representatives start adjusting existing customer orders and placing new ones for the next four weeks. Notice we are talking about making promises now: since we no longer have any stocks to ship from, we are now making promises based on our capacity instead of actual stocks. The first two weeks quickly fill up, and the sales team forecasts that the following two weeks will be sold out as well, but it currently considers being a bit more conservative about the following four weeks.
In our example, we established that toilet paper was manufactured on three production lines, consisting of 8 operations on 7 work centers. Let say each production line has its first 6 work centers dedicated, but the slicing and packaging one is shared among all lines. Every 6h, a batch of toilet paper can come out of a production line’s WC#6 and takes a 30 minutes tour on the roller before reaching the global slicer.
Since we are limited by raw material, equipment won’t be used at full capacity in the first four weeks. Line B will make 2 batches of 3-ply, then 1 of 2-ply, line C will make 3 batches of 2-ply, and line D will make 3 batches of 2-ply at partial capacity (21,100 kg).
Figure 5: Load Profile of the slicer (colors based on source product line)
We have no overload here since the slicer has much more capacity than the three production lines it serves. It is still useful to see the available time if we need to schedule maintenance activities.
Now consider there is an unexpected 12h downtime on the slicer on Monday after the first batches. It would impact the production for the rest of the day and even the following day. If we cannot afford to build more than one batch of work-in-progress inventory on each line while this downtime is addressed, there would be an overload. This overload couldn’t be fully processed on Monday, and would thus be taking some free space early on Tuesday morning, but would also push back a bit the packaging of Tuesday’s first batch. With the CRP, once we can estimate how long the downtime will be, jobs can be quickly rearranged so that the different teams could all push back their work of 8h for the 3rd batches on Monday in order to avoid building up too much work-in-progress inventory (because 2nd batches are already started when the downtime occurs, they will have to stop with a work-in-progress inventory of large rolls at the end of the rollers). The production target for Monday wouldn’t be reached, but the shop floor could catch up on Tuesday. Customers, which are groceries and department stores, would have to be notified to mitigate the effects of the backorders. For example, they could be restricting sales to end-users to only 1 package by person – if they aren’t already doing so!
Notice we referred many times to customers in this blog post about capacity. Customer experience can be greatly enhanced when production planners have an accurate and real-time view of the shop floor activities. They can then share the real-time analytics with the sales team, who can manage customer expectations accordingly.
Let’s bring even more complexity here. Imagine you have 2 instances of slicers: one for both line B and C, and one for line D, with downtimes happening randomly (this plant’s workers desperately need training to perform change-overs without breaking anything!). Can one quickly optimize production activities at any time to route rolls either on their targeted instance of the slicer (preferably) or on the only slicer instance that is up and running without a graphical view?
Instead of spitting more load profiles tables in this post, let’s have a look at a cloud-based CRP tool in action! Gain insights on how a CRP tool can efficiently schedule and tweak production efficiently by investing some time watching the Rootstock webinette on production planning (8 minutes demo, unfortunately not on toilet paper topic)! Pay attention to the work center group view: it’s one great view allowing you to balance the load among multiple work centers with the same expertise or machine if there is any – and it’s a possible answer to this last question on the slicers’ random downtimes!
Finally, if you would like us to assist you with your capacity management challenges, on Rootstock or elsewhere, let us know, and we’ll be glad to help!
************
To read all posts in this 4-part series on Capacity Planning check out: