At the end of the last posting the statement was made that it is necessary to address green manufacturing and the role of digital enterprises in the context of informing the customers each enterprise serves as well as those to whom the enterprise appears as the customer. Importantly, this has to be done both external to the organization as well as inside. What does that mean? With apologies to again referring to past postings … this starts with the Google earth view of manufacturing. This was first introduced in a posting way back in 2009. It bears refreshing everyone’s memory! The idea, shown in the image below, envisions one starting at the enterprise level and then zooming in to increasingly detailed parts of the manufacturing enterprise spanning
To affect process control at the process level (here for a metal cutting operation - hence the reference to chips and cutting tool) one needs to have data and response at the micro/millisecond rate. As one moves higher in the structure the timing scale slows down proportionally. When reaching the enterprise level where supply chain management and asset management is of concern the decision and response time is longer. This is not, however, to imply that there are not decisions in supply chain management that do not occur more rapidly in some cases. In fact, the interesting thing about this type of representation is to look for the dependencies at lower levels on decisions and responses at higher levels. For example, a catastrophic tool failure at the process level that causes substantial down time and loss of availability of the machine and line could ripple up to the production planning and scheduling level if the disruption is substantial.
Each interface between the different levels in the manufacturing enterprise, and their accompanying data rates, decision rates and response rates offers an opportunity to add noise to the system (meaning reducing the reliability of the data or, at least, increasing variability in the data) and must be accounted for.
OK … but the Google earth view above has another side to it on the right. In this figure the potential range of influence and impact of the social effects (or dimensions) of the manufacturing enterprise are illustrated. These are challenging to represent in a simple drawing like this but will affect, at the process level in the facility, mostly the workers and support staff (for example, working conditions, safety, training, pay, etc.). The potential range of impact expands as one moves up the levels culminating with the supply chain which can have a national, regional effect or, within that country, a specific community (for example, air quality, water quality, healthcare and education, etc.). The “data rates” for acquiring impacts for these impacts (assuming one can quantify them sufficiently) will be similar as on the “hardware” side although may not be at the fastest level. Monitoring working conditions for exposure to chemicals or other contaminants for the worker at the machine may need to be done on a second or faster rate. Data on health care or educational levels of workers in the supply chain will be less frequent.
Never-the-less, is one desires to apply the digital enterprise concepts to tracking the sustainability of enterprise operations from energy and resources to social concerns the data challenges will be impressive.
We need to get more specific. What will this data look like? To illustrate, three examples are presented her derived from recent research at the Laboratory for Manufacturing and Sustainability (LMAS) at Berkeley. They address a facility level view, a line level view and a machine level view. They all concern manufacturing that centers on production of machinery using a range of processes but with a strong component of machining and metals fabrication. Energy consumption is a common metric here as it is readily measurable.
Facility level - The first example is drawn from the research of Dr. Nancy Diaz (N. Diaz, “Development of Energy Models for Production Processes and Systems to Inform Environmentally-Benign Decision-Making,” Ph.D. Thesis, University of California, Mechanical Engineering, 2013) and focuses on the comparison of electrical energy intensity (kWh/meter squared/year) for four production facilities of a major Japanese machine tool manufacturer. The data reflects the consumption of energy by the machinery in the plant, the heating, ventilation and air conditioning (HVAC) and lighting. The four plants address different parts of machine tool manufacturing from ballscrew production, the most precise (and hence requiring most exacting control of the environment - temperature and humidity) to less demanding machining and assembly. The figure below, from Diaz thesis, illustrates the dramatic range of consumption of energy per unit of floor area for the different factory functions and energy uses. The ball screw production facility has the highest HVAC energy intensity since these are exceptionally precise components that determine the quality and eventual performance of the machine tool and must be produced under the most stringent environmental conditions. Ball screws are turned by the motors on each linear axis of the machine and cause the table on which the workpiece is mounted during machining to move under the control of the computer program. They define the precision and accuracy of the machine movement (to a great extent).
What do we learn from this? First of all, at this level, it is clear what the “relative cost” of different manufacturing processes are in terms of energy (and likely other resources) … precision is the highest due to the requirements of the facility, quality of the consumables, etc. Think of the semiconductor industry as at the high end. Then, it also shows where the greatest potential for improvement is in the process efficiency to reduce this intensity. But, it is also necessary to determine the total impact , meaning, the measured intensity (energy / unit area) times the total area involved in this type of production. If this is a small part of the total production then it might not be the first priority. If it is a major component then it could offer big improvement.
Systems/line level - The second example is drawn from the research of Dr. Stefanie Robinson (S. Robinson, "An Environmental and Economic Trade-Off Analysis of Manufacturing Process Chains to Inform Decision Making for Sustainability," Ph.D. Thesis, University of California, Mechanical Engineering, 2013) and focuses on the energy and resource consumption in a process chain with the objective of establishing a basis for trading off the potential for upgrading specific operations in the line. This was based on research conducted with a major heavy equipment manufacturer in the US. The figure below shows a schematic of a multiple operation process chain and a detail of one of the process operations with typical input and output of energy and other consumables along with waste and emissions.
With a representation of a process chain, and the individual operations, one then needs to determine the consumption and rate of outflow of the major consumables and waste streams. One can appreciate that it is necessary to do a rather detailed analysis of the inputs and outputs (wasted and worn tooling, scrap from production, leakages, etc.) to be accurate. With this data, the actual resource consumption and associated economic and environmental cost can be determined. Then, the impact of changes in any of the production steps can be evaluated both in terms of productivity and quality as well as environmental (energy, global warming, water) effects and associated costs as shown below.
System consumption metrics and environmental and economic "cost"
Process level - This third and last example bores in more finely on the process level detail for a machining operation. Data from this level of analysis would feed into the systems/line level just described. This example also draws on the work of Dr. Nancy Diaz in the above cited thesis. This work developed a generic method for calculating energy consumption during a realistic machining operation on a precision milling machine based on constant and variable contributions of the material removal rate (MRR). The MRR is a driver of productivity in a machining operation and is based on real time data of the feed rate of the cutting tool, the cutting speed and the depth or width of engagement in the case of milling studied here. This data is now available in real time from the machine controller thanks to standardized interfaces and data formats such as MTConnect and associated software. It is also available from the numerical control program driving the machining operation (that is the path the cutting tool takes in the machining operation) but that is often inaccurate due to actual the performance of the machine in operation. The curve below shows the specific energy (Joules/cubic millimeter, J/mm3) as a function of MRR.
The significance of this data is that the designer or production engineer can determine the energy to create a part feature from knowledge (either estimated from the tool path or measured in process). And, this was determined for a variety of machining conditions with different tooling - so it has some breadth of application. For other materials, however, the curve would likely shift up (if a more challenging material to machine - more energy per unit of material removed) or down (if easier to machine).
So this is all driven by data! Lot’s of it collected at different speeds and representing different “views” of the enterprise. It is encouraging but humbling. Fortunately, as referred to in the previous blog posting on the digital revolution, communication speeds, computational capability and speed and the hardware spitting out the data from machines and systems are more common, less expensive and more reliable. The expression “drinking from a firehose” comes to mind!
But, the good news is that no one will be thirsty! Some of the tools for using this data in productive and green operations will be covered in the future.