Green Manufacturing and Resiliency

What's resilience?

This week the discussion is on "resiliency". And, how it relates to manufacturing and, in particular, green and sustainable manufacturing.

But first, a final comment on leveraging (the subject of the last three posts). In a discussion about leveraging with some of my researchers last week it was suggested that, in fact, leveraging works in both directions - from manufacturing towards the product and from manufacturing back to material selection. We'd been discussing the "forward" direction with respect to changes in the manufacturing process that may require some investment of resources (or energy, materials, etc.) but which will yield a substantially larger reduction in life cycle impact of the product in use and, hence, a good 'return on the investment.'

The "backward" look is equally sensible but I don't have an immediate example in mind but, when I do, it will be the subject of another posting. Here, we can make decisions in the product design or manufacturing that influences material selection. For example, we can choose to use a production technology that is, perhaps, more energy intensive but allows us to choose from a wider range of materials including some that are less energy intensive to produce (lower embedded energy), less hazardous or better for operation of the product to reduce impact.

That is, we can mirror leveraging in both directions about the manufacturing process. And, interestingly, this could make our systems more reliable and resistant to disruption due to, say, materials shortages or other disruptions due to impacts.

This is a great lead in to our discussion here - resiliency.

The dictionary (Merriam-Webster on-line) defines resiliency as "1: the capability of a strained body to recover its size and shape after deformation caused especially by compressive stress or 2: an ability to recover from or adjust easily to misfortune or change" (and they give the example of "emotional resiliency"). The second definition is probably closest to what interests us here - recovering from unexpected or unwanted change or misfortune. Think supply chain disruption due to, for example, floods in Thailand or earthquakes in Japan.

Actually, we can characterize these disruptions in terms of our ability to foresee or predict the disruption or plan for it. Things like earthquakes are unpredictable. You can choose not to build your factory in an earthquake zone (but some choose not to worry about that if you can build the structure "resiliently"). You can't always predict or anticipate other system stressors like labor disruptions, mineral or material shortages, equipment malfunction, etc. But you can try to take steps to reduce the impact (or inoculate your system from their effects). Planning, redundancy, alternate sources, careful choice of components/suppliers/sources, etc. all can help.

If you "Google" the term 'manufacturing resiliency' you will get a number of postings and articles dealing with reducing downtime due to disasters and other unanticipated events that result in reduced employee productivity, revenue loss, damaged corporate reputation and missed service levels. These "unanticipated events" can be caused by power outages, natural disasters, or other disruptions to a manufacturers’ supply chains and critical material or part suppliers.

Of course, many suggest that IT is the solution … more information faster means fewer surprises. Maybe.

Others suggest that a cause of concern is the volatility of prices in the materials/metals markets. A recent article by consultants KMPG titled "Global Metals Outlook: Manufacturing Resilience" discusses this in some detail. These are not manufacturers - but metals processors and suppliers - the folks that provide materials to manufacturers. Logically, their strategies include cost optimization, trying to gain more control over raw materials and, interestingly, locating assets closer to customers or suppliers. The report states "More than one-half (53 percent) of respondents from metals companies say their organizations are considering localizing or customizing operations to improve the efficiency of their supply
chain, compared with 43 percent of manufacturing companies more widely. Given the size and bulk of their products, shipping costs are a major concern."

Interestingly, the report did not mention anything about helping their customers make better use (increased yield) from materials or lengthening the product life cycle to better control demand. Honestly, most of the experts interviewed in this report were not the operating engineers but from the financial and management side. So that is not a big surprise. But, that would work!

Back to resilience. An excellent review of "resilience thinking" is in Ecology and Society in a 2010 paper reviewing resilience as part of adaptability and transformability - all key aspects of the dynamics and development of complex social-ecological systems. We're going to dive into social metrics and manufacturing at some time in the future but, for now, keep it close to engineering. From the paper cited above, we see that "Resilience was originally introduced by Holling (1973) as a concept to help understand the capacity of ecosystems with alternative attractors to persist in the original state subject to perturbations… In some fields the term resilience has been technically used in a narrow sense to refer to the return rate to equilibrium upon a perturbation (called engineering resilience by Holling in 1996)."

Hollings wrote a foundational paper on resiliency (the full cite is Holling, CS (1973) Resilience and Stability of Ecological Systems, AnnualReview of Ecology and Systematics, 4:1–23.) In this paper Hollings discussed the difference between engineering resilience and ecological resilience. He considered that the engineering system has one equilibrium state only, while the ecological system has more than one equilibrium state.

So, simply put, resiliency is the ability of a system (say a supply chain or production system) to return to a stable operable state in the presence of "attractors" (or in engineering terms, disruptions) that would tend to move the system into another state of operation - presumably less stable, or less profitable, or less environmentally benign.

It is not too hard to see where risk comes into this and, if the risk is induced by unexpected events (like floods) the resilience of the system will be the ability of the system to return to normalcy with the least disruption. And, with respect to "equilibrium states" it is clear that manufacturing systems may have many (since they have many different components) and it might be preferable to move to a new equilibrium state if it can be shown that it is more green or sustainable!

So, let's draw the conversation back to manufacturing. Equilibrium is a very well understood engineering term and refers to a state of rest or a natural condition that a system will revert to when left alone. In the case of manufacturing, say a production system, equilibrium might be when the system is operating as designed with the requisite result or output. A complex supply chain might be said to be at equilibrium not when it is stopped or doing nothing (as in the engineering definition "state of rest") but when it is functioning smoothly. I realize this is not a precise definition but it will suffice for our discussion of resilience here.

I recently was exposed to the use of resilience with respect to green manufacturing and sustainability in the context of the National Institute of Standards (NIST) use of the term as part of a description of their sustainable manufacturing program. The site explains that "the sustainable manufacturing program will enable advanced manufacturing processes that include new manufacturing methodologies, manufacturing information systems, and effective industry standards. The Program results will advance U.S. leadership in sustainable manufacturing, resulting in technologies that support the application of Key Performance Indicators (KPI’s) to access and decide on production networks which require much less energy and materials, reduced waste and optimal logistics. By using these technologies industries are ideally positioned to optimize their processes and maximize their efficiency and resilience."

Lot's there - methodologies/technologies, information systems, key performance indicators (KPI's), standards - all with the purpose of helping to make decisions on production processes and networks that use less energy and materials, reduced waste and optimal logistics. And, hence, make the processes and networks more resilient!

Let's continue with how that might work in practice next time.

Leveraging Manufacturing, Part 3

The big finish!

That's a pun - gear finishing, leveraging, get it?! OK - blogger's license.

We will finish up our example of leveraging with this post. Although there was a long dead space in postings, recall that the example was from a recent paper from our research group and focussed on  the gear train as used in transportation. The premise was that the surface finish of gears contribute substantially to the efficiency of power transmission. Better surface finish yields better efficiency.

It was described that the gear manufacturing process chain is relatively complex with several options available to the manufacturer at each fabrication stage. In this example it is assumed here that the main process chain would be unchanged and that only gear finishing would need to be altered to produce gears with higher surface finish. For reference, the full citation to the paper on which this series is based is “Evaluating the relationship between use phase environmental impacts and manufacturing process precision,” CIRP Annals, 60, 1, 2011, pp. 49-52. I'll send you a copy if you want one.

The "leveraging" comes in with the expected fuel savings due to the better efficiency of the gear operation due to the better surface finish. We need to determine if the increased consumption of energy in finishing is paid back in the improvement in the operation of the gear train and accompanying reduction in fuel use. And a result of reduced consumption of fuel in the auto use phase we see reduced global warming potential (both from the reduced fuel used and the avoided impact of producing the fuel.)

Using the basic approach outlined in the last post, it was first necessary to determine the 'cost' of manufacturing improvements relative to surface creation. We do this by looking at the specific energy consumption requirements of the grinding process used in this part of the manufacturing process chain. From published data, for example from Professor Tim Gutowski at MIT, we know that the specific energy (meaning the amount of energy to remove a volume of material) for a grinding process assumed to be reflective of standard automotive gear finishing applications is about 200,000 Joules/cm3 for a process with a removal rate of about .01 cm3/sec. So, in English, if you want to remove a cm3 of material at this rate it will "cost" you 200KJ.

Using this approximation and the relationship between surface roughness and removal rate from earlier researchers we are able to estimate the increased specific energy required to decrease the surface roughness of the final gear drive reduction relative to the representative gear finishing process. This  estimate provides an upper bound to the manufacturing energy usage - meaning it should not exceed that since it is a convective estimate. Primary energy (energy needed for either the manufacturing process or moving the automobile) demand for the process and GWP emissions were then determined assuming a Michigan electricity mix (7015.2Btu/kWh and 0.7131kg CO2-eq/kWh, respectively. We assumed we were manufacturing the auto in Michigan.

The figure below shows the increase in PE demand and GWP emissions from electricity usage
in the manufacturing phase due to decreased surface roughness. This means, as we put more

energy into the grinding process to improve the surface roughness (recall, smaller is better in surface roughness) there will be a corresponding increase in global warming potential (GWP). Lower primary energy consumption is better for a given set of process conditions. In the figure we see two curves, one for the least sensitive relationship between process removal rate (x = 0.60) and the other for the most sensitive (x = 0.15). This shows the change (improvement) in surface roughness one can achieve by "spending" process energy - reducing surface roughness from the nominal by 50%, for example, will cost us 1.25MMBTU. (Read the graph as the x-axis at 100% is the typical surface roughness and moving towards 0 indicates reduced roughness or better surface.

Now to the automobile's primary energy consumption based on gear train efficiency. The fuel consumption of a vehicle is dependent on the power that the powertrain must deliver to meet the commanded acceleration while powering any accessories (e.g. air conditioning) and overcoming losses in the drivetrain and engine. Because this analysis considered only changes to the drivetrain efficiency, the power required for any accessories and frictional losses in the engine were neglected since neither would be affected.

The U.S. EPA Federal Test Procedure 75 (or FTP-75) emissions driving cycle was used to represent a standard driving scenario for this analysis. The decrease in fuel requirements was calculated for each surface roughness, Rq, of the gear pair in the final drive reduction. The resulting decrease in energy that must be provided by the fuel was then determined by integrating the decrease in fuel power. All deceleration events were removed from this calculation since a deceleration event does not require power from the engine. Modern engines are operated to fully combust fuel, and so the PE demand and GWP emissions were determined assuming that the fuel source was regular, unleaded gasoline (1184.8Btu/MJ used fuel and 0.0948kg CO2-eq/MJ used fuel, respectively.

The figure below details the relationship between the surface finish (stated in Rq, microns) of the gears in the final automotive drive reduction and the reduction in automotive primary energy demand (gas!) and the comparable reduction in global warming potential.

This figure shows that decreasing surface roughness (Rq) lowers PE demand relative to a standard finished final drive reduction from 2-5MMBtu depending on the operating temperature, To. The earlier figure showed that a 20-60% reduction in roughness increases PE demand in the manufacturing phase by less than 0.5MMBtu. Comparing these analyses indicates that improving the manufacturing precision of the final drive reduction can provide a substantial reduction in the life cycle impacts of an automobile. Since the final drive reduction is one of several gear pairs in a vehicle, the impact of manufacturing precision on the entire vehicle drivetrain could be much greater.

This analysis showed that a relationship exists between the manufactured precision of a product and its environmental impacts over its entire life cycle. In the case of automotive drivetrain components, this relationship was found to be positive. However, it may not be true for every product and is largely dependent on the intended function of the product. Ultimately, if a manufacturer is concerned with environmental impact when considering a process or system design, then he should improve the manufacturing precision if the resources required for the improvement are less than the potential benefit of the improvement in the use phase of the manufactured product.

This is summarized in the figure below. The figure plots surface roughness (to the right is rougher) and the comparable primary energy demand difference between the use (auto operation) and manufacture (creating the surface by grinding).

You can see that, for the standard gear finishing operation at the right we set the difference (cost minus savings) at 0. Then, according to our analysis improved surface roughness, even though it costs something in the manufacturing phase, yields a good return (savings greater than cost - so negative delta) over a wide range of surface roughness (and corresponding process conditions). If we push it too far and try to get too fine a surface finish (going far to the left in the plot), the manufacturing energy needed outweighs the benefits in improved performance - it costs too much to do. The trick is, first, finding the relationship the allows us to define this curve and, second, determining when the lower limit is reached and it is no longer "environmentally profitable" to improve the process further.

Clearly there are things, some more important than others, that we are leaving out of this analysis. But, it is a pretty good, and accurate, example of leveraging. For example, we should measure other aspects of gear finishing processes so we can include other environmental impacts such as water, industrial fluid, and raw material usage. We might also consider other manufacturing and product effects such as increased or altered process consumables for the manufacturing process, will this more aggressive finishing process result in decreased process yield (that is, more rejects) and what impact does this change have on the service life of product. These could be additional benefits as well as offer some disadvantages.

I encourage you to read the paper if you want the full details. There is tremendous potential in this approach.

Finally, we just hosted at Berkeley the 19th CIRP Life Cycle Engineering Conference. We had almost 180 participants from all over the world and it was a great series of presentations and discussions on many aspect of life cycle engineering as it applies to manufacturing. The "theme" of the conference was "Leveraging Technology for a Sustainable World." You can read a short overview of the conference in a blog posting on the BERC blog space prepared by one of our lab members, Katie McKinstry. Katie's posting is titled: GLOBAL ENGINEERING CONFERENCE SHOWCASES SUSTAINABLE MANUFACTURING INITIATIVES 28 May 2012 | BERC News. Enjoy!