Dynamic Adaptive Scheduling – Is it Time To Move Beyond Make-to-Stock versus Make-to-Order?

This article is based on a presentation I made recently to the annual conference of SAPICs, the professional body for Supply Chain Management in South Africa.  The June 2022 SAPICS conference was the first in-person event since the COVID pandemic.  It was great to be back, working face-to-face with the supply chain community.  Much of the discussion was around planning methods, particularly DDMRP, and how well they can respond to the current challenges: Supply chain shortages, rocketing input costs and gathering storm clouds over the world economy.

I hope that this article can provide some ideas around how to respond to a rapidly changing environment.  It may be that we need to move past the “Make-to-Stock” versus “Make-to-Order” models for manufacturing.  Even the periodic re-positioning of buffers in DDMRP may not respond fast enough to a mix of changing market conditions and customer needs.

A few years back, before DDMRP, I implemented a replenishment-based planning system for a manufacturer of safety footwear. We set buffer levels for finished stocks, and the system monitors inventory levels, on a daily basis, and as soon as a replenishment signal popped up, we would whack it with production order.  

Whack-a-Mole and the problem of chasing demand with production orders.

A bit like the game of “whack a mole”. The trouble was that there were so many past due sales orders and low stocks, that before long, pretty much every mole had popped up!  “Never mind”, we said, “When we have the right buffers in place, the replenishment system will work just fine.”

Guess what?  Several years later they were still in the same position and focused on planning each sales order. The replenishment system had failed.

I used to ask clients, “are you a make-to-order or make-to-stock company?”. The answer would often be “well …………”. This got me to thinking about the mindsets behind these two examples.

This is small town USA in 1930’s, a typical stockist with his stock all around him. He knows that his customers expect to walk out of his shop carrying their purchases, and if he doesn’t have it in stock, he loses the sale.

This is another kind of retailer, and this one I used in the 1960’s. He also has stock in his fridge, but there is a manufacturing process, and his demand is very visible, he can see the length of the queue, so how many chips to put in the fryer, but he also calls out to the customers in the queue. “Cod, haddock, plaice? How many?” His customers are standing in front of him, he knows what they want, and that drives his manufacturing process. He is a make-to-order operation.

More recently we were approached by a food company, who asked the question “How much stock, of each SKU, should we be holding?” They used the term “Model Stock”.

“Great!” we thought- “We can use some DDMRP thinking to size the buffers for them.”

 So we had the company extract transaction-level detail from there ERP system, for one year, and they sent us over three million records to analyze.

The findings were:

A very steep Pareto (as with most companies). The low volume products are made intermittently and supplied from stock. The typical saw tooth pattern. The most popular product has it’s own dedicated production line, which runs every day. The other large volume products are produced each day, or alternate days.
The average customer order is for delivery 9 days ahead. Very few require immediate or next day delivery.
The average delivery is made 1 day early, but up to 9 days. That’s your buffer right there.

Demand is fairly even throughout the month, and days of the week.

Daily production is covered by open sales orders 95% of the time.

Most “stock” is not stock, it is product on its way to a customer. But scheduling was based on replenishment of total stock, not just “free” stock.

And they were asking “how much stock should we hold?” Our answer was “None”, most of the time. In reality, they were a make-to-order company, stuck in a make-to-stock mindset.

Just like the fish and chip shop owner, they have a very clear picture of customer demand. The customers is standing in front of them, telling them what they need.

Part of the problem, again in common with a lot of companies, is that you have a production planner who has visibility of the finished stocks and is driven by a replenishment philosophy. His job is to get the stuff made and put it in the warehouse. Then it is the job of the dispatch planner to get it to the customer. He has visibility of customer demand but doesn’t look at planned production – what’s coming. There is a brick wall between the two, they have quite independent functions.

When you remove the brick wall wonderful things can happen.

At a SAPICS conference long ago, a US speaker came up with this slide that said  

“The best kind of warehouse is no warehouse at all, and the best substitute for a warehouse is information.”

I was very impressed with this idea, and shortly afterward a client of mine, a Durban based automotive filter manufacturer was planning to build a new, bigger warehouse. The biggest market is Johannesburg, and about 60% of the volume went to their Joburg depot. When you went to the yard, the Joburg truck was always there, being loaded. When it left, the next empty trailer would arrive. However, there business process was to put production into the Durban warehouse, register the stock movement in their ERP system, then figure out where to send it. So, 60% of the volume came off the end of the production line, into the Durban warehouse, waited there for the planning process to catch up, then onto the Johannesburg truck. And they needed a bigger warehouse in Durban.

Come on! The solution is pretty obvious, isn’t it? Why not move it straight from the end of the production line, and onto the truck? One of the excuses was “We have to put it in the warehouse before the system will print the dispatch notes”.

So, we got the two planners talking to each other, and got the planning system to print pallet labels, ahead of each production run, that read ….. Joburg, Joburg, Durban, Joburg, Joburg, Cape Town, Joburg, Bloemfontein …… etc. the fork truck driver then put the pallets on the appropriate waiting trailer. Warehouses on wheels. To fool the ERP system they did a virtual, not a physical, stock movement transaction, in and out of the warehouse.

Right, let’s get back to the food company. At one end of the scale, there are small volume products, where one hour of production is a month of demand. As you are producing the product, you don’t know who the customer is going to be. They are always going to be managed on a make-to-stock replenishment basis. At the other end of the scale, there is the most popular product, with a dedicated production line, customers who place orders well in advance, and will accept delivery early. As each pallet comes off the end of the production line, it is already assigned to a customer order. They will always be managed on a make-to-order basis. The production schedule will be driven from the order book. If you are getting too far ahead into your order book, you may decide to shut the line for a while. If you are making orders that are due today, or even past due, then you may schedule some overtime to get ahead.

I have been asked, “What about customers who want immediate delivery? Don’t you have to hold stock for them?” Well, no, you dynamically re-assign scheduled production to those orders.

But what about those products in the middle, that lie in between the two extremes? Some are going to flip-flop between make-to-order and make-to-stock in response to changing patterns of supply and demand. There may be temporary capacity constraints or even raw material supply issues that erode your ability to maintain the buffers.

So, when the stocks are low, and the crocodiles are snapping at your feet, you look beyond the replenishment signals and ask “Who is the customer?”, “Will they accept a smaller quantity for now?”, “Are they OK with a delivery next week?”. Customer visibility is key, but when the buffers are back in place, then you can go back to focusing on replenishment.

This is a Tesla Model 3. It has dynamic adaptive suspension.
When it detects that the road is a bit bumpy, it lifts, and when the road is smooth once more, it lowers.
Now this is a dynamic adaptive scheduling system. It works in a similar way and here the road is smooth. You can see that assembly line 1 will be working on two works orders (manufacturing orders, production orders) this morning, and then there are replenishments, order recommendations, in DDMRP speak. The 3rd item there needs to be blessed with a works order. Thursday is a public holiday, and we are not working the weekend, so we are looking at the schedule till Wednesday next week.
Now I change a parameter in the system to look further ahead for any sales orders that are not covered by stock. Now see what happens when the road gets bumpy.
After the two works orders, it becomes a make-to-order schedule for two orders, then lapses back to being a make-to-stock schedule. It’s saying, these two sales orders are due Monday next week. Let’s keep the customer happy before we continue with replenishing our buffers.

I’m going to be going through this tool on a webinar, 30 June 2022.  If you would like to attend please register here.  Anyone who attends will get a copy of the dynamic scheduling system example that I show you here.

https://us02web.zoom.us/webinar/register/1916558267909/WN_UWggedmhQueoC9ITmtD17Q

Leave a Comment

Scroll to Top