In this short article I would like talk about a few aspects of electrical energy consumption in industry and potential saving strategies. First, I will outline some global challenges we’re facing and motivations for reduction of primary energy use. Then I will focus on two approaches that help to improve energy efficiency in the enterprise sector. I will discuss a case of a part loading robot in discrete manufacturing plant and show possible improvement supported by experimental data.
So what’s with this energy thing?
Nowadays progressively more attention is being paid to energy efficiency and energy use in general by the enterprises. Why is it so? Well, there are two main drivers.The first one is environmental. You already know that high energy demand equals is tight to high CO2 emissions, as most of our primary energy production is still based on fossil fuels (nearly 44% in EU but more tha 80% globally). Nuclear power and renewable also carry a certain environmental impact related to e.g. nuclear waste management or production of PV panels, therefore reducing energy usage is environmentally relevant no matter which energy source we are using.
Second reason for implementing energy saving measures is quite obvious energy costs money. EU average price of kWh for enterprise users averages at around 12 euro cents which is a few cents more that of China. To remain competitive with the fierce competition from the east Asia western countries usually go for wide automation of processes to improve work efficiency. Such type production is, however, more energy demanding than the one where manual labor was once utilized. Sophisticated, robotized machining centers have incomparably higher energy demand that simple, manually operating lathes or milling machines. As a result energy is becoming increasingly important part of the monthly bill of the enterprises. As the relative energy cost per product is rising the incentive to reduce it becomes more visible and therefore increasing energy efficiency is gradually becoming one of the key factors of future of manufacturing.
Is there anything that can save us?
Sure there is. Efforts to improve the way we manage our energy are being taken by the scientific community and an increasing number of enterprises willing to cut down on their monthly bills. So how do they do it?
Energy saving measures in industrial sector usually can be divided into two main approaches. First one is related to using more efficient equipment, such as high efficiency motors, variable speed drives, state of the art chillers systems or simply whole new machines which are consuming less than their obsolete predecessors. This approach is effective, yet it requires investment and obviously your return time may vary from a year to a few years, depending on the particular case. Everyone who has ever bought an energy LED light bulb for his home understands the pains paying today and waiting for the money to come back over the prolonged period of time.
The other approach that may bring significant reductions is implementing energy management policies. Basically, it is about using the energy in a smarter fashion. Recalling a household situation, having a good habit of switching light when you live the room, defrosting food in the fridge, turning off the water while brushing your teeth can be a good, small scale example of such policy. Today I would like to focus my attention on this second approach and show how it can be implemented in a manufacturing plant.
Now something real – an industrial robot case
Let’s consider a particular case of an industrial robot, whose job is to load parts to a CNC transfer machine with multiple machining stations. I will demonstrate that even with a heavily loaded system there is still some room for energy-oriented thinking and savings are indeed possible.
Having said that, let’s take a closer look on our scenario. Imagine a production cell which consists of a robot and a CNC machine. Whenever a workpiece machining is finished robots picks it up from the machine and lays it on the outcoming conveyor belt. Then it picks up a new, raw part from the incoming conveyor belt and inserts it into machine, just like in the picture below.Machine that we consider is a four station, three module transfer machine with rotary table. This means that there are three independent stations where part is being machined and one station for part handling (inserting raw and taking out completed pieces). A key characteristic of a transfer machine that it will be machining three parts simultaneously which radically increases the throughput. A good example of such type of processing can be seen in the video of a multi-center transfer machine embedded below (start at 2:15).
Look at the picture below. Every part that flows though our cells undergoes a certain, 5 cycle process. In the first cycle it will be inserted by the robot into Station 4. Table will rotate 90 degrees and three sequential machining cycles will be performed, each followed by a table rotation, moving the part to the next position (Stations 1-3). In the final cycle (no. 5) part will land again in the Station 4 where it will be retrieved by the robot.
Let us take a closer look on the activity of the robot that is handling the raw and completed pieces. The action of swapping raw and completed parts takes a fraction of the time of the whole cycle (the red squares in the picture above) and for the rest of the time the robot remains idle, as it must wait for the all three modules to complete machining.
Noticing this fact is an entry point to our energy analysis. Below I attached a response surface that show how motion parameters, namely target axis velocity and acceleration affect consumption of the energy. The black circles “floating” on the top of the surface are points collected through three phase power measurements at the input of the robot. These measurement points where used to create an mathematical model of the energy demand of the DC motors of the robot. They include static friction, velocity-dependent losses (vicious friction and iron-losses) and copper losses. If you are interested in the energy modelling and identification techniques I used, I invite you to read my publication dedicated to this particular topic.
We can easily notice that the faster the target velocity, the higher the energy consumption. Also high acceleration will result in increased losses but this factor doesn’t seem to have a dominant role here. We can see that lowering the velocity nearly to 0 will allow tosave up to 60% of energy per motion.
But how slow can we really go and how much energy can be truly saved? Well, it all depends on the cycle time. To have a clearer vision on the problem I used a numerical optimization tool to find what is a minimum energy (with respect to motion velocity and acceleration) that can be achieved for a given motion time.
We can see that the value is on the left side bounded by the maximum allowed speed limit that the robot cannot exceed. As we increase the motion time the energy decreases reaching an asymptotic value at around 0.9Wh. This value indicated the minimum energy that we can theoretically spend performing that motion.
Hey, but what about the machining process? Aren’t we forgetting about something? Well, the robot can be slowed down but not more than to the duration of the machining cycle, as it would lead to machine blocking and eventually loosing throughput. Therefore we can “stretch” robot’s motion to be Just-in-time before the table rotates. Look at the figure below, which depicts this policy. Knowing the current robot motion time and machining cycle time will allow us to predict total energy saving, which you can read from the plot.
Summing it up
I hope I was able to demonstrate that with this straightforward methodology a quick, energy-efficient solution can be found and painlessly applied to this particular type of problem. What is more it haven’t required practically any material cost. But, will it work in any case and everywhere? Not necessarily, you have to know your system. It this particular case the cycle time of the machine defines production rate, therefore robot speed can be easily adjusted without interfering with the process itself. With different type of system in which changing the speed will affect the throughput, most likely more sophisticated techniques will needed. Often to find energy optimum, without risking any blocking or starvation of the machines in the line, an energy aware system balancing procedure shall be utilized. But that is a topic for a different article!
I would like gratefully thank European Commission for founding my industrial research on the work Initial Training Network “EMVeM” (GA 315967)