Tuesday, July 13, 2010

Degrees of Perfection, Part 2

Part 2 of a series

The degree of perfection discussion in the last posting was centered on the term "buy to fly" ratio popular in the aerospace industry to indicate material utilization. I stated that we need to consider all the peripheral "stuff" associated with a product like electronics, appliances, clothing, food, etc. which usually comes packaged so we might want to consider a sort of "buy to fly" ratio for conventional products.

I am aiming in this series to get to a more engineering discussion of exergy (or available energy and useful work) to address this. But, I want to play with this  more fascinating buy to fly concept for manufacturing a bit more.

In fact, based on a number of comments I've received on this, others also are intrigued by the extension of buy to fly to more general manufacturing applications and processes. Ralph Resnick, an old friend from my early days of chasing burrs, now at NCDMM, suggested something along the lines of "energy to manufacturing" for tracking the useful output of the process for the energy input. So, let's explore some other ways to implement this idea.

Last week I attended a research review conference held at a machine tool builder's product design and development facility in Northern California (DTL/Mori Seiki). We toured the facility and I noticed a machine, the Mori Seiki NT1000 mill turn center, that touted it's abilty to provide the same functionality in a 95 x 106 inch (or 2.4 x 2.7 meter) footprint that other machines requiring twice the size deliver. That is, more output per unit of floor space occupied. This measure is traditionally emphasized in the semiconductor industry where space in high tech clean rooms is very expensive.

You might recall that some time ago (last December to be exact), as part of a discussion about ways to green machines and processes I did a virtual comparison of a set of individual machines versus a multi-function machine. This NT1000 machine is one of those. So, in addition to the efficiencies of eliminating the other standalone machines, the reduction in floor space gives extra benefit that can be measured in terms of plant environment, lighting, construction costs and materials, etc.

But, let's push this a little further. The NT1000 and similar machines by other manufacturers has an approximate volume of 15.5 meters cubed and a work volume of approximately 0.06 meters cubed - a ratio of almost 260 to 1. I was curious how this compared to machine tools in general meaning - do we always need that big of a machine to make small parts? (The NT1000 is designed for precision machining for medical devices, automotive hardware, watches, instrumentation, etc.)

A few years back I had a visitor in my laboratory from Doshisha University in Japan. Professor Hirogaki was working on "downsizing" machine tools and presented some interesting data on what is "typical" in the machine tool industry - but he measured the relationship between the weight ratio (machine weight to removal weight) as a function of removal weight (or mass actually). The figure below, from Professor Hiragaki, shows some typical results (again you'll need to click on this for details).

Here, the "target" is a weight ratio of 1 which we approach as the machine size increases. So this would suggest that bigger machines are closer to "perfection."

Interestingly, if we plot the similar ratio for the multi-function machine we've been discussing, the data fits this graph nicely (down in the lower left of the x-axis). The machine mass is given as 8000 kg and the equivalent steel workpiece volume (removal volume) is about 470 kg for a ratio of machine mass to work volume mass of 17. But, and this is a big but, the multi-function machine replaces about 3 equivalently sized machine tools. So, by this "buy to fly" comparison - it looks quite good.

Others are working in micro-sized machines to make micro-sized parts to address this "why do we need a big machine for small parts?" issue. I was curious about how they match up. One leading company, Microlution sells a machine (the 363S CNC 3 axis horizontal mill) with a working volume 2x2x2 in (or 5x5x5 cm - roughly) in a machine volume of 24x24x54 in (or 61x61x137 cm). Volumetrically, this yields a ratio of machine volume to work volume of, gulp, almost 3900.  I did not do the mass ratio to see where this fits into Professor Hirogaki's curve.

Recall that the "conventional" multi-function machine tool above had a volume ratio of 260 to 1.

So, there are limits to using these type of calculations perhaps. Trying to make machines the size of the work volume (the "target" ratio of 1 in Hirogaki's figure) may not be feasible for small footprint machines. The trick is ... how to make larger parts with small features on the small machines?

Finally, lest we beat up on ourselves in the machining business too much, let's look again at microelectronics. In 2002, researchers Eric Williams and colleagues published a paper in Environmental Science and Technology on the "1.7kg microchip: energy and material use in the production of semiconductor devices" The chip, a 32MB DRAM chip with a mass of about 2 grams, requires a total weight of secondary fossil fuel and chemical inputs to produce it and use it  estimated to be 1600 g and 72 g, respectively. This is a buy to fly ratio coming of 835.  They also consider the use of water and elemental gases (mainly N2) in the fabrication stage which are 32000 and 700 g per chip, respectively. Using only those water and gas fabrication numbers gives us a buy to fly ratio of 8175 - a new high!

This points out the need for considering some additional ways to measure our "degree of perfection." That is the perfect transition to our discussion of exergy next time in part 3 of this series!

By the way, this posting is our one year anniversary of the blog. Happy Anniversary! We started this blog last July 15th and, thanks to your reading and feedback, it has been a great year. Thanks for following!


  1. Happy Anniversary, indeed! The techy stuff is the best way to celebrate my favorite blog.

    If we could standardize by volume the process by which inputs, energy included, are transformed into outputs, the prefect shape would be a cylinder, where all the outputs are useful, for nature, for humans or for both. The current processes are truncated cones with different "buy-to-fly" ratios symbolized by the ratio between the two bases. The cylinder's ratio is the perfect 1, no volume is lost.

    How would you include the degree of specialization of a machine as another criterion for measuring the degree of perfection? In order to minimize waste, including idle time, it is important for the machine to be flexible, adaptable to changes in demand, capable of processing large and small product quantities. Since both relative and absolute input/output values are relevant, if a machine or production line requires very large quantities to be efficient (as is the case of the semiconductor industry with its highly specialized machines and automated lines) does that increase or reduce the degree of perfection?

    There must be a trade-off between downsizing and keeping the machine as universal as possible. How can this trade-off be estimated?