A Guide to Implementing the Theory of
More On Buffers
Why, why, why, are the buffers in drum-buffer-rope so commonly mis-understood? Well, I will go on to suggest when we return to the main page that it is in part the way we “draw” time on our diagrams. But that isn’t much of a reason – it sounds more like a poor excuse. Here are some better reasons.
Deming illustrated this phenomenon so well with Rule 4 of Lloyd Nelson’s Funnel Experiment – a random walk, or “worker training worker (1).” Or in this instance “expert” training “novice.” When the expert isn’t actually expert we are in for one hell of a problem.
There are far too many places – especially on the internet and internet list groups – where people claim to show or discuss buffering correctly, and it is not correct at all.
How can we blame the novice? How is a novice even to know? The brief answer is that the novice isn’t to know, but that doesn’t mean the novice is blame free. The onus is on the novice to check with the original source material to make sure. The most accessible and authoritative source is Schragenheim and Dettmer’s book Warp Speed (2). Read it! Better still, buy a copy of Goldratt’s production simulator and do it (3)!
The erroneous approach – that the buffer is a pile of work in front of the constraint – is entrenched and continually re-taught even though it is incorrect. That it is so easily accepted as correct is because it is so closely conforms to what we do now in most situations.
Think about it. If we have a constraint in a system, what is the chance that it is starved of work? Most often it is not starved of work at all, indeed most often work is continuing to build in front of it – although whether it is working on the work in front of it or not is an entirely different matter.
Of course in the initial stages every part of the process may similarly have no lack of work to work on. Nevertheless it makes common sense that the weakest link should have work waiting for it if all the other links have more capacity and are working to their best efforts. So, it does not seem at all nonsensical that the buffer is indeed this work. And this is where the error starts.
Think the following through.
If we had a pile of work in front of the constraint before we implemented drum-buffer-rope and we have a pile of work in front of the constraint after we implement drum-buffer-rope then what in then has buffering achieved? Nothing!
Hold on to that thought.
It might appear that the pile of work has achieved something, the constraint is doing more now than it used to do and we don’t want to starve it. But did we starve it before? Probably not. Are we starving it now? Not very likely! In fact we are now critically aware of it’s importance. So the pile of work in front of the constraint is not aiding the exploitation of the constraint. So what is it doing?
Do you know why we don’t starve the constraint once we begin to exploit it? Even though the amount of material demanded by the constraint increases? It is because there is always “something else” in the pile that can be “pulled forward” and put through even if the right material according to the production plan is missing! This is the key.
Drum-buffer-rope began in make-to-order environments. Environments where there is a commitment to make a delivery on a specific day or a specific hour of a day. Environments where timeliness is important. This is what the buffer protects. It protects the timeliness of the system.
It protects the timeliness of the system by subordinating the raw material gating and all other steps up to the buffer origin so that the material arrives in good time to be processes at the planned time so that it can eventually be shipped at the planned time.
This is what the buffer does.
Buffering Is Subordination – Buffering Is Not Exploitation
Our current knowledge causes us to be side-tracked into believing throughput or material output is the key outcome of drum-buffer-rope; we make more stuff that is worth more income. However, if we do this with continued lousy delivery times then we won’t achieve anything. We must deliver better throughput and better timeliness if we are to sell more now and in the future.
Making sure the constraint is working to the best of our ability is exploitation. Having a big pile of stuff in front of the constraint has no bearing on this. Making sure that we have the right stuff in the right place at the right time is subordination. This does have a direct bearing on what the constraint can do and when.
That is why the buffer in drum-buffer-rope is concerned with time and not quantity.
(1) Neave, H. R., (1990) The Deming dimension. SPC Press, Inc., pg 98.
(2) Schragenheim, E., and Dettmer, H. W., (2000) Manufacturing at warp speed: optimizing supply chain financial performance. The St. Lucie Press, 342 pp.
(3) Goldratt, E. M., (2003) Production the TOC way (revised edition). North River Press.
This Webpage Copyright © 2006-2009 by Dr K. J. Youngman