Thursday, September 22, 2011

Breakthrough in Material Planning: Demand Driven MRP

For the first time in over 30 years, Material Requirements Planning (MRP), is undergoing a fundamental improvement. A major new development, dubbed Demand Driven MRP (DDMRP) is moving from theory to practice, and the results are impressive. If you care about manufacturing ERP, you would be wise to pay attention.

Carol Ptak recently called my attention to her work with Chad Smith at the Demand Driven Institute, which they founded in 2010 to promote the concepts of DDMRP. I've known Carol for many years, through her work as past President and CEO of APICS and her time at PeopleSoft, where she was an early proponent of making MRP more "demand driven." I wrote a brief blog post in 2003 covering this subject (see: PeopleSoft Strengthens Its Manufacturing Offerings by Acquiring Demand Flow).

So, I jumped at the opportunity to set up a briefing on DDMRP with Carol and Chad for me and my associates Bob Gilson and Nick Hann at Strativa.

Bringing MRP into the 21st Century

Before we look at some of the key concepts of DDMRP, let's review the history of material planning.

MRP, first developed in limited fashion in the 1950s and 60s, really took off in the 1970s, when computer systems enabled widespread adoption and APICS undertook the "MRP Crusade" to popularize it. It was a great advance over previous material planning techniques, such as statistical order point (popularized during the Second World War), which viewed each inventory item separately. The big conceptual breakthrough of MRP was to separate dependent demand (sub-assemblies and purchased items) from independent demand (e.g. finished goods and service parts). MRP, therefore, provided a holistic, or system-wide view of inventory.

MRP (material requirements planning) morphed in the late 1970s and 80s into MRP II (manufacturing resource planning) and ultimately ERP (enterprise resource planning). But the heart of today's ERP systems (at least in the manufacturing sector) is still the MRP processing logic that is essentially unchanged since the 1970s.

In practice, MRP relied heavily on demand forecasts to drive planning and used safety stock inventory to cover variability in lead-times and forecast errors. The results, though far better than the old order point systems, were often excess inventory and less-than-acceptable customer service levels.

Just-in-Time (JIT) inventory and lean manufacturing techniques were, in part, a reaction to the complexity of MRP and a desire to obtain better outcomes. Introduced in the 1980s, JIT was a simplification in material planning and it took inspiration from the quality management movement and Toyota Production System in Japan. JIT is essentially a "pull system"--relying upon simple demand signals, such as Kanbans, from customers to suppliers up and down the supply chain, often with little or no computerized support. Unlike MRP, which viewed inventory as an asset, JIT viewed inventory as a "waste" and sought to minimize it wherever possible by minimizing variation in supply and demand, and reducing setup times to enable smaller lot sizes. But its emphasis on inventory reduction, lack of a system-wide view of inventory, and incomplete planning equation created brittle supply chains, subject to disruptions.

Embracing and Extending MRP and JIT

Demand Driven MRP is not a completely new method: it builds upon and extends the concepts of MRP while borrowing the best features of lean manufacturing. Like lean manufacturing, it seeks to "align efforts and resources as close as possible to actual demand" (a so-called pull system) while at the same time, like MRP, provide "visibility to the total requirements and status picture across the enterprise."

The authors' background in the Theory of Constraints is evident. They are not looking to compromise between MRP and lean manufacturing. Rather they recognize and seek to satisfy the legitimate objectives of both. For example:
  1. The greatest extension of DDMRP is the introduction of supply chain modeling prior to generating material plans, as shown in steps 1-3 in the Figure at the top of this post. Here, the organization determines the optimum places where inventory should be held. This is a step that MRP simply does not address. MRP and even Advanced Planning systems (APS) generally take inventory stocking points and safety stock levels as givens and plans within them. At the opposite end of the spectrum, JIT techniques are blind to the overall supply chain. Each node of a pure pull system is only sensitive to the demand at the next downstream operation. DDMRP, on the other hand, models where inventory should be held in order to minimize lead times and reduce variability where it matters the most.
  2. In terms of inventory, DDMRP stakes out a middle ground between MRP and lean manufacturing. It does not view inventory as a waste, as lean manufacturing does with its goal of "zero inventory," and it does not seek to establish safety stock levels in a static way, as MRP generally does. Rather it seeks to hold the right amount of inventory at the right place in the supply chain "to promote flow but minimize working capital," and "to size and dynamically adjust those strategic stock positions" based on a set of rules dominated by six factors.
  3. DDMRP deals with lead times in a more realistic fashion than traditional MRP, which in calculating manufacturing lead time assumes all components are in stock, or in calculating cumulative lead times assumes nothing is in stock. Neither are good assumptions in most environments today. DDMRP introduces a concept it calls Actively Synchronized Replenishment (ASR) Lead Time (ASRLT), which represents "the longest unprotected sequences in the bill of material" where "protection" is defined by strategic stocking points. These points decouple, compress, and ultimately define the calculated lead time of an item.
  4. MRP is only a planning tool and JIT is only an execution tool, whereas DDMRP is both a planning tool (in the modeling and planning stages) and an execution tool, in the execution stage (see Figure at top of this post).
  5. DDMRP greatly reduces the emphasis on accurate forecasts in driving supply plans. Demand is driven entirely or largely by actual customer demand (typically sales orders), which can then be satisfied from compressed lead time due to the strategically placed inventories at the subassembly or component level.
These are just a few of the key concepts of DDMRP. For a more complete view, see the additional resources listed at the end of this post.

If there is any doubt that DDMRP is a major breakthrough, consider this: the book everyone considers the "Bible" of MRP was written in 1974 by the late Joseph Orlicky, and the second edition was authored in 1994 by the late George Plossl. McGraw-Hill, publisher of Orlicky's Material Requirements Planning has just released the third edition, which is authored by Carol Ptak and Chad Smith, the brains behind DDMRP. This third edition now incorporates the concepts of DDMRP, building upon the work done by Orlicky and Plossl, two of the fathers of MRP.

Proof in the Pudding

Think DDMRP is just a nice theory? Not so. The benefits have already been demonstrated by at least two early adopters:
  1. Oregon Freeze Dry, after adopting DDMRP, saw a 20% increase in sales 20% and 60% inventory reduction in one division, along with 60% reduction in make-to-order lead-time and 20% inventory reduction in another division.
  2. LeTourneau Technologies (LTI) is an interesting case study in the natural resources sector. The firm implemented DDMRP in one of its two plants, while letting the other plant continue with traditional MRP. Both plants had similar resources, bills of material, and material planning personnel.

    During a boom and decline cycle of 2005 to 2008, the DDRMP plant grew revenues from $270M to over $620M, while growing inventory by only about $80M. In contrast, the traditional MRP plant experienced the similar growth in revenues, but inventories grew at the same pace. When the recession hit in 2008, the DDMRP plant , however, was well positioned with lean inventories while the traditional MRP plant was exposed to a huge amount of inventory liability.
Additional case studies should begin to appear as other organizations gain experience with these new methods.

What's Next?

I believe that enterprise software vendors, especially those focused on supply chain management, are going to move quickly to begin to incorporate DDMRP concepts into their systems. So far, there is only one software provider that has done so, Replenishment+® from Demand Driven Technologies, of which Chad Smith is a paid advisor. However, I see indications that some other vendors are moving in this direction. In other words, I do not believe availability of software is going to be an obstacle.

If there is any obstacle to wholesale adoption of DDMRP, it is going to be in the general level of resource planning skills in many manufacturing organizations. The concepts behind DDMRP are not simple. Many practitioners tasked with responsibility for MRP systems today do not have a deep understanding of MRP principles, even of traditional MRP circa 1974. How are they going to grasp the concepts behind DDMRP? I think this will be the key limitation hindering widespread adoption, unless manufacturing organizations are willing to re-invest in professional development in a sustained way. Conceptual education--not just software training--is going to be key.

A corollary observation is this: production and material planning is going to become an even more critical function for manufacturing and distribution firms. It always has been, of course, but it will become even more critical in industries where some players have the skills to adopt DDMRP and others don't. No longer just a back-office function, resource management should once again become an inviting career path for young people.

Additional Resources

There is much more to DDRMP than I can outline here, such as its ability to accommodate seasonality, ramp up/down in production, and end-of-life scenarios, all of which are troublesome for traditional MRP systems and especially lean-manufacturing systems. Therefore, it is best to point readers to the following resources.
  • Demand Driven MRP. The flagship DDMRP website maintained by Carol and Chad, with a good introduction to DDRMP. Free white papers, videos, and podcasts are available on a number of DDMRP topics.
  • Demand Driven Institute. The educational and consulting organization promoting the concepts of DDMRP. The textbook I mentioned earlier, Orlicky's Material Requirements Planning, Third Edition, is also available here, with a supplemental DVD.
  • Orlicky's MRP. This is the official website of the new edition of Orlicky's Material Requirements Planning.
There are several white papers and videos linked at these sites that go into more depth on DDMRP. Training classes are now being rolled out. In addition, the authors will be presenting more on this subject at the APICS International Conference in Pittsburgh, October 23-35. If you are planning to be at this conference you will be wise to register for this session, as it will probably be standing room only.

Related Posts

APICS Returns to Its Roots
PeopleSoft Strengthens Its Manufacturing Offerings by Acquiring Demand Flow
Lean Manufacturing: Not a Complete Solution without Information Technology
Lean Manufacturing Doesn't Really Need Software, But Software Can Help
Lean Thinking is Still More Than Software


Matthew King said...

Hi Frank,

I wonder if this has much in common with Demand Sensing systems.

DDMRP sounds promising though, particularly if it's easier to implement and operate than Advanced Planning Systems.

Frank Scavo said...

Hello Matthew. I checked the link on Demand Sensing systems, and the very first sentence reads, "Demand Sensing is a next generation forecasting method ..."

Based on that, I would say DDMRP has nothing in common, as one goal of DDMRP is to eliminate or at least greatly reduce the need for forecasts. As I understand it, DDMRP does this by strategically placing inventory at intermediate points in the supply chain so that delivery can be accommodated within customer lead time. (I'm sure I am over-simplifying, so see the links I reference in the main post).

Whether DDMRP is easier to implement or operate than APS, well, I'll defer to someone who has experienced both. As I indicated in the main post, I am concerned that there is quite a bit of conceptual education required to be successful with DDMRP, and I'm not sure how many organizations or materials planning professionals are willing to invest in that much education. I think, however, where the business case justifies it, it is a very worthwhile investment.

Matthew King said...

Hi Frank,

Thanks for setting me straight...

I should have written: "I wonder if Demand Sensing and DDMRP have anything in common".

After closer inspection, the Wikipedia article is not great. It doesn’t even mention “Pattern Recognition” which is central to the technique. This link is better:

So Demand Sensing on its own, is a forecasting method geared toward short term planning, theory being that medium term forecasting is typically a fruitless exercise - an assumption in common with DDMRP.

Component four of DDMRP encompasses the use of short term forecasts - or customer orders - or both. Demand Sensing slots into this component. In other words Demand Sensing doesn't compete with DDMRP, but can instead form a part of component four.

I'm happy with that conclusion - because I can stop thinking about Demand Sensing and start thinking about DDMRP in terms of the bigger replenishment planning picture.

In regards to DDMRP, I watched these two clips:

The second clip mentions how the Replenishment+ application is connected to an execution system - filling me with the same horror as APS. But it appears to seriously empower the planner, unlike APS which tends to put the keys in the hands of the information technology professional.

There's a lot for the planner to digest, so I agree with you about the education barrier. But maybe the smart planners will finally have a tool that actually meets their needs – in a sophisticated yet practical way - allowing them to extricate themselves from their beloved spreadsheets more often.

clive boulton said...

Hi Frank,

DDMRP seems more like a rework of the planner action report, adding strategic inventory buffers and alerts. Compounding things Replenishment+® extracts transforms and loads back the data into a host ERP system (a necessary evil). On other vendors, I learned SAP designers in Walldorf are working like Billio to transform MRP to take advantage of HANA's predictive analytics. I actually think SAP and Replenishment+® have lots to learn from Amazon, Nike and others applying predictive productivity commerce and distributed big data to supply chains. I can't understand why the ERP vendors are not thinking from the ground up.