Quick Response Manufacturing (QRM) is touted as the answer to Lean for hi-mix, lo-volume environments typically encountered in SMEs like job shops or machine builders. Personally, after having read the book and some papers about QRM, I got interested in QRM for the management of make-to-order products along make-to-stock fast movers, typically managed with the more traditional, anonymous or product-specific kanban (see also an earlier post on pull upstream and downstream op the order penetration point). So when Rajan Suri, Mr. QRM himself, was in the Netherlands last week, I took the opportunity and enrolled in a two-day workshop on QRM with him. Expectations were high, but although QRM does propose some interesting tools, overall it was somewhat disappointing. A post on what to expect and what not to expect from QRM.
QRM: Time as a Competitive Weapon
In times where low-cost countries like China and India have proven to be massive economic powers, SMEs in many countries are looking for ways to compete. Suri proposes time as the key to their competitiveness. This is at the core of his approach called Quick Response Manufacturing or QRM. QRM is about lead time reduction. And it focuses on SMEs characterized by low volume and a high variety of (possibly) custom-engineered products. Because of the lead time reduction, so Suri argues, a lot of waste and therefore overhead and indirect cost typically attributed to the product cost are reduced, thereby reducing the standard cost of a product.
In this respect, QRM does not propose anything new. Time based approaches already existed well before QRM. Lean as well as TOC in general have lead time as a main target. Just remember Taiichi Ohno saying: “All we are doing is looking at the time line, from the moment the customer gives us an order to the point when we collect the cash. And we are reducing the time line by reducing the non-value adding wastes” as recorded in the 1988 book on the Toyota Production System. And in this respect, I also strongly recommend Goldratt’s “The Race” from 1986 that very well explains the power of time, as QRM calls it. So QRM and Lean share this common target through which waste can be eliminated and avoided.
The thing less to like in the way Suri positions QRM is that he states that Lean doesn’t work in the typical job shop environment. This is somewhat simplistic I think. Just remember that Toyota in fact exactly developed their production system to cope with lower volumes and a more proliferated product portfolio compared to US car manufacturers. I think it is important to note that Suri in fact speaks about the traditional kanban system not being applicable rather than Lean as a whole. By speaking of Lean as a whole he fails to recognize that Lean is far more than kanban and that even JIT involves a lot more than kanban (think of heijunka, the full work system and the work order sequence system). But we’ll come back to that.
MCT: Visualizing Lead Time
As part of QRM’s approach, Suri proposes to focus primarily on one key indicator of performance and progress: time. This measure is operationalized using a metric called Manufacturing Critical-path Time or MCT. The good thing about MCT is the fact that it includes the administrative process preceding the actual production process so often encountered in make-to-order or even engineer- or configure-to-order environments.
Another thing I liked about it was the way lead time is visualized. Unlike time in a VSM, the MCT shows waiting time proportionally to the time in which the part is actually touched. For Lean initiatives focused on reducing lead time this proportionality aspect in fact helps teams better focus on the problem of waiting time. Furthermore, it allows you to visualize multiple feeder lines, which isn’t that easy in a VSM. But then again, a VSM isn’t always focused on just the time aspect either. From the example shown here, you can maybe see that the MCT map looks a bit like an inverted Gantt chart.
The funny thing is that this MCT map made me think about the kind of time-based process maps we used to make in the Business Process Re-engineering (BPR) era. I even found a nice one from one of the projects I was involved in, in the mid-90’s, focusing on administrative lead times (sorry, this one is in Dutch).
As you may have noticed, I mentioned the MCT approach distinguishes between waiting time (what it refers to as white space) and touch time (so called grey space, in which someone works on the job). It is important to note that this isn’t the same as value-add and non value-add as for instance setup and material handling are seen as touch time in the MCT. The reason Suri mentioned for this were related to managing change: because these times are typically relatively short compared to the white space, calling them non value-add may put off employees to further work on lead time reduction.
The thing which again is a pity, is that although Suri mentioned that the MCT map can be complimentary to the VSM, he also introduced the MCT metric as being simpler and superior to Lean’s measurement of the seven wastes. This is silly, as I never even have seen anyone proposing the seven waste categories as actual indicators. In fact, Lean’s process efficiency indicator is very similar to the MCT metric comparing grey space to white space.
Cells: the Product Family Approach
The next thing proposed by QRM, is to structure your organization around so-called Focused Target Market Segments (FTMS) that are at the basis of the creation of QRM cells. To be honest, this is almost identical to the analysis you typically perform when preparing for a Value Stream Analysis, preparing the implementation of kanban loops and developing a Value Stream based organization.
Furthermore, the additional elements required to form QRM cells are comparable to the Lean approach to organizing your plant, i.e., develop autonomy and team ownership for decision-making in the cell, multi-skilling and focus on continuous improvement, particularly focused on lead time reduction.
But is everything interpreted in exactly the same way? No. Suri provided an example of team ownership whereby an operator from a cell went out to collect his own parts from a paternoster system. Now although this might be interpreted as team ownership, when your focus is on reducing the lead time of your products, I personally don’t think having operators go out of the line to collect their parts is such a good idea…
Also here, Suri tried to differentiate Lean and QRM by stating the Lean cells only know unidirectional flows and are inflexible. I think this view is based upon a simplified view of a Lean organizational structure. Not everything in a Lean organization is (yet) organized into the well-known U-cells with dedicated, right-sized machines focusing on only a small product family. But this doesn’t mean Lean doesn’t consider any other form of organizing yourself around flows. Again, a pity, as QRM’s second core concept is extremely similar to the approaches promoted in Lean thinking.
When discussing this topic, I also asked my main question for the workshop: what does QRM propose as an approach when a cell includes both a small set of make-to-stock fast movers and a larger set of make-to-order slow movers and more exotic parts? To my disappointment, the answer was to split these cells into two cells and to dedicate different resources to them. Now although I agree on the principle in the long run, not all companies can simply duplicate their resources and dedicate their work centers to different families just like that.
System Dynamics: Good Ole’ Queuing Theory
QRM’s third core concept is referred to as “System Dynamics”. Nice wording maybe, but nothing more than queuing theory with a new label. Suri thereby does reference John Little and his well-known Little’s law (from 1961), however, he unfortunately does not give credit to John Kingman and his Kingman formula (also from 1961) upon which QRM builds heavily.
Coincidentally, I recently published a blog post on Kingman and his formula from queuing theory that you can find here: http://dumontis.com/2016/07/muri-mura-kingman/
So unfortunately, also here, QRM doesn’t really propose anything new, other than a new vocabulary for already existing theory. The thing which is good though, is that Suri brings Little, but also Kingman in particular, back into focus. As the literature on Lean doesn’t do a good job in explicitly referencing queuing theory in its writings and concepts.
POLCA: Generic Kanban for Managing FIFO Lanes
A larger part of the workshop was dedicated to the POLCA system. POLCA stands for Paired-cell Overlapping Loops of Cards with Authorization. Without going into too much detail here, POLCA is a card based system (like the kanban system is), that links two cells by a number of cards in a loop. A POLCA card is returned to its origin cell when the destination cell has finished work, in order to signal that is has free capacity. To contrast this with the traditional, anonymous and part-specific kanban card, a POLCA card therefore is a capacity signal.
By having POLCA cards in a loop between two cells, you in fact keep the work content between the two cells constant. POLCA therefore is a way to implement the same thing as kanban tries to do, viz., keeping work-in-process (WIP) stable and thereby lead times in control (the Constant WIP or CONWIP principle also known from factory physics and even Goldratt’s Drum-Buffer-Rope approach, based upon Little’s Law from 1961). Furthermore, by releasing constant work packages (in terms of required capacity) instead of just work orders, QRM just as much applies the principles behind levelling (heijunka) as Lean tries to do.
With POLCA as a release signal based upon available capacity and with the purpose of keeping WIP in control in a certain part of your value stream, you of course still need a sequenced list of work orders that can then be released. In QRM, such lists are called authorization lists and each cell needs one that is regularly updated when changes in priorities occur. To create such lists, QRM proposes the use of MRP at the value stream level (so cells are seen as “black boxes”) which is similar to the hi-level use of MRP together with flat bill-of-material (BOM) structures in Lean. As QRM combines pull for execution with MRP for the authorization lists, Suri speaks of POLCA as a push-pull system, to differentiate it from pure schedule-driven push and anonymous, part-specific pull systems. I think this is not a correct way to represent POLCA, as to me it still is a pull system focusing on keeping WIP in control by tying the release decision to the actual WIP in the value stream.
So the underlying concept is not really new, but is the technique? I am sorry to say that even the technique proposed by POLCA isn’t really new. As early as 1982, Masuyama of Toyota’s production control department already described Toyota’s work order sequence system, and in the same year Hayashida described the sequenced pull and broadcasting system in use with Toyota’s suppliers. Hall described Yamaha’s PYMAC system, also known as Synchro-MRP in 1986, using kanban to pull from a sequence list generated by MRP. Later, Sandras (1989) used the term “generic kanban” to do the same as the POLCA cards. And Hopp and Spearman called it the “job kanban card” (1996). Although the exact details of each of these techniques sometimes differ, their purpose and even their elements are fairly similar: a card that is used as a release signal upstream and a sequenced list based upon which the operation then subsequently actually releases the next work order into the flow.
In its most simple form we know this as visual management of FIFO-lanes, whereby the preceding operation monitors its outgoing FIFO-lane per destination and when a position is free, he can take the next work order from upstream with that same destination (looking at the routing on the order sheet typically accompanying the parts in the FIFO-lane). The maximum size of the FIFO-lane in fact also keeps WIP between two cells in control, just like the POLCA or generic kanban cards. Now I do think working with cards is somewhat more robust than only working with visual management of the FIFO-lanes, particularly when the layout doesn’t really facilitate visual management of your queues. But when possible, visual management is of course far more simple than cards.
QRM: Not New, But Some Nice Additions
Coming back to my initial question: what’s new about Quick Response Manufacturing or QRM? As you have seen, not that much really. The focus on lead time reduction, the cell approach to organizational structure, the use of queuing theory (or system dynamics) and even POLCA are not new concepts or even totally new techniques.
The main thing which is a nice addition, I personally think, is the visual representation of lead time using the MCT map. That can help teams and companies be more conscious about the waiting time in their system, even beyond the shop floor. Another thing is the POLCA system. Although not really new, it does represent a very practical way, and described in detail, to actually implement the CONWIP principle in a lo-volume, hi-mix environment.
Overall, QRM has taken a subset of Lean elements, relabeled them under the umbrella of QRM and targeted SMEs characterized by a lo-volume and hi-mix of (possibly customized) products. Can these companies benefit from QRM? Sure they can, as the core concepts of QRM as well as its tools are effective ways to improve the operations of these companies, no doubt. But to position it as different from Lean to me is just a marketing fad. There is nothing really new under the sun, but hey, in the end it doesn’t need to be new to be effective. But let’s not create the illusion QRM is different from Lean. It isn’t.
And as for my question on mixing make-to-order with make-to-stock items in the same cell: QRM doesn’t have an answer for this. So back to the gemba it is, working with the teams to discover new territory and continually improving the ways that we currently already apply in mixing and effectively managing these types of products in our cells, with the aim to control and reduce our lead times and reliably serve our customers.