All improvement is change, but not all change is an improvement. We can optimise sales, or marketing, or manufacturing, or supply chain, or HR, or IT. Indeed, you can optimise any subsystem of your organisation. But how will it affect your business outcome?
[ Listen to audio version, read by David Hodes]
To optimise the system constraint we have to take a step down from the lofty heights of theory and operating philosophy and engage with the suite of the three pragmatic methodologies Goldratt invented, covering the requirements of projects, production and supply chain replenishment. These three approaches are sometimes called the “proven solutions” as there is an abundance of evidence that they deliver results which would otherwise seem unreasonable and impossible. The three are Critical Chain Project Management (CCPM), Drum-Buffer-Rope (DBR) production management and Dynamic Buffer Management (DBM) for the supply chain replenishment solution.
“When you focus on the constraint,
you can improve the system as a whole”
The full power of CCPM as an optimisation engine becomes apparent when you have to solve the complexity of running multiple projects off a single resource pool. We typically find these types of environment in, amongst others, product development businesses, consultancies, engineering functions and businesses and software developers and integrators. But, before going into optimisation for the portfolio, we should have a quick think about the individual project.
What prevents a project from being completed in no time and at no cost? Sound like an absurd question? Not if you’re looking for the constraint. What constitutes the time dimension? It would have to be the duration of the longest sequence of tasks that cover all the required scope, from initiation to completion. This longest sequence is the idea behind the concept of the critical path. And, what of the zero-cost part? Unless your project has free labour and materials, you’re unlikely to get away with zero cost. So, to come as close as you can to the ultimately optimised project, you have to consider what the tasks are, what sequence they’re in, who’s going to do them and what materials they’re going to use.
Suppose you make the simplifying assumption that there are no spatial or access constraints to any given work front and that there is no uncertainty associated with the known timing of any of the tasks. In that case, you can derive a deterministic lead time and expected completion date for your project. But, life’s not like that, is it? Who has an infinite number of resources and doesn’t have to deal with contention between them? And whether it’s a physical production space or access to a unit of software code, it’s infrequent that queues don’t form behind an active packet of work.
“Who has an infinite number of resources and
doesn’t have to deal with contention between them?”
CCPM does the job of optimisation by considering the longest sequence of tasks after levelling the load across all tasks, given the number of resources you make available to the project. The CCPM method also addresses the issue of task uncertainty by reducing the duration of each task, followed by aggregating and storing their individually held contingency in carefully placed buffers. At the end of the planning process, we have an optimised schedule, which explicitly takes into account both resource availability and task uncertainty.
But, planning is for getting you into things – you have to get yourself out. We cannot say that we have genuinely optimised a project until the method used for execution is itself optimal. In execution, the CCPM method calls for the measurement of progress along the critical chain relative to the amount of buffer, that is, the aggregate contingency consumed. Monitoring this critical ratio provides project managers with the wherewithal to focus their efforts on those few or even single task which needs management attention and prioritisation of resource allocation.
In the multi-project environment, we once again predicate optimisation on finding the constraint. Imagine we have prepared a portfolio of projects using the CCPM method for single projects. We then stack them up, one above the other, all starting on the same day. We measure in daily vertical slices what the load is by specific resource type and compare that to the capacity we have available.
On each day, there will be one resource type whose ratio of load to capacity is higher than any other. The load will be less than the available capacity for some of those resource types, whereas, for others, the opposite will be true. In simple terms, based on what your budget allows, given a relevant time horizon, one resource type will be more loaded than any other.
To optimise, we then have a choice. We can either accept our resource constraints and extend the overall duration of the portfolio by shuffling the pipeline of projects sufficiently to the future to eliminate the resource constraints. Or, we can add additional resources until we eliminate the given bottleneck and discover what the next one is, as it emerges from our calculations.
Once we have optimised the pipeline in the planning phase, we once again use the signalling system as per the case of the single project outlined above. Depending on the value of the critical ratio, that is chain completed versus buffer consumed, we can plot all of our projects on a single control chart to visualise which project most urgently requires focus.
Since it is axiomatic that constraints govern the rate of portfolio completion, it is also the case then that non-constraints have capacity available. We therefore achieve optimisation when unconstrained resources subordinate to the needs of the constraint. But, that’s a topic for our next article on the third focusing step – Collaborate around the proposition that the constraint is the rate-determining step in the creation of value.
“Planning is for getting you into things—
you have to get yourself out”
We achieve optimisation in the production domain through the application of the Drum Buffer Rope (DBR) scheduling system. As the name infers, the drum, also known as the critically constrained resource (CCR) beats the rhythm of the production system and is akin to the concept of takt time in lean. In the perfect world, the rate of production at the drum exactly equals the rate of sales, with single-piece flow and no batching. The buffer is a store of inventory just upstream of the drum, placed to prevent any shocks to material flows from common cause variation. The rope is how we send a signal for the release of materials such that those materials arrive at the drum just in time for their scheduled operation at the CCR.
The DBR method is the TOC way to accomplish what operations people call finite scheduling. It’s difficult to imagine arriving at an optimum schedule without taking on the significant challenge of finite scheduling. As with CCPM, the first order of business is to level the load – what lean practitioners call heijunka. Typically you take all of your confirmed production orders for the next week or fortnight, depending on your scheduling horizon, and calculate the demand on each resource type.
For example, if you have production orders for several different models of widget, each of which takes varying amounts of time on lathes, grinders and welding, you would calculate the total time by work centre and find which one is the most constrained. That work centre then becomes the drum. The drum schedule is the very detailed, hour by hour schedule that allocates work to the drum such that the production system delivers the optimum outcome in terms of on-time performance.
In managing execution, we use buffer management to determine where we need to focus such that we have a very reliable drum schedule. In brief, to run buffer management, we split the time from material release to the allocated time for commencement of drum production into three, creating a green, orange and red zone. As the date with production draws closer, the planned work moves from the green to the orange and finally to the red zone.
“It’s difficult to imagine arriving at an optimum schedule without taking on the significant challenge of finite scheduling”
Once the planned work is in the red zone, you would hope that you would have met all conditions necessary and sufficient for production to start. These conditions might include materials, tooling, drawings, standard operating procedures, safety measures and the like. Buffer management allows you to make visible, ahead of time, any missing production requirements and to focus management attention on expediting their supply.
An additional component of the DBR system for optimisation is how we build continuous improvement into the system. If, for example, a particular production batch misses its scheduled slot, you can record the reason. After a while, you’ll have a statistically significant sample of reasons for non-adherence to schedule. We can arrange these reasons into a Pareto chart, and improvement efforts would then naturally focus on the most common cause for slippage.
The optimised supply chain is capable of squaring the circle of minimising investment in inventory while at the same time maximising customer service. These two goals appear to be at odds with each other unless you understand the principles of what drives the amount of inventory you must carry to achieve a given level of due date performance.
The TOC replenishment solution provides remarkable results, but as ever, it is counter-intuitive. Ask most people, and they will argue that we must keep as much inventory as close to the customer as possible. This strategy would, however, send us broke. Imagine a shoe shop having to stock every style, colour and size of shoe just in case someone walks in to buy a pair?
“The mathematics of the central limit theorem
comes to our rescue”
What factors then make for the optimum mix? It turns out that the amount of inventory we have to carry is a function of how long it takes to generate an order, how long to complete production, how long to ship and what the forecast demand is for the product. Further considerations include an understanding of the variability fo each of those metrics and the crucial concept of the tolerance of the customer for delay in instant gratification.
The variability part of the equation is quite discombobulating if we are to make accurate predictions for one product at a time. However, the closer we get to statistically meaningful sample size, the more the mathematics of the central limit theorem comes to our rescue. Simply put, the larger the sample size, the more predictable the aggregated result.
So, where is the place of highest aggregation? Think of a distribution system as a V, with the point of supply at the bottom and the consumption points at the top. The deeper you go down the V, the more aggregation there is. Thus, counter-intuitively, if you hold most of your inventory at the bottom of the V, you will be better prepared to address the volatility at the top. But, only if you place a great deal of emphasis on the velocity of your product from the bottom to the top.
The key to this optimisation idea is to reduce batching to an absolute minimum: purchase order batching, production order batching, and shipping order batching. And understand your customer. If I wear a size 13 shoe, I’m a lot more likely to be willing to wait for exactly the style I want than someone whose feet are right in the middle of the bell curve.
There is much to be said about how the three proven solutions from TOC, CCPM, DBR and DBM can help optimise your business. They have proven capable when competently applied, to time and again deliver not only outstanding results but also the means to have those results continuously improve.
This is Part 4 of our series on The 5-Step FOCUS.
» Learn how our our services can help you.
» Return to view all articles.
» SUBSCRIBE TO NEWSLETTER and never miss an article
[Background image: Mountain climbers, Charlie Hammond on Unsplash]
Constraints aren’t your enemy—they’re your friend. If you want the most bang-for-buck for your effort, despite the connotation of the word constraint, it’s precisely at that place you’ll find the leverage you’re looking for.(more…)
In a recent workshop, I asked the team to state their goal. The number they gave me—expressed as a volume of safe, reliable tonnes—was a good deal less than the design capacity of the plant. A little surprised, I asked them what was going on.(more…)
Discover better ways to do better work.
We alternate our own actionable articles with three relevant links from other authorities.We’ll only use your email address for this newsletter. No sales calls