Industrial energy efficiency potentials: an assessment of three different robot concepts

Abstract

The rise in energy consumption and the associated costs instigate financial concerns among industrial energy consumers. For industrial processes addressing heating and cooling as well as material transformation, a wide range of energy efficiency measures have been developed and successfully implemented. In contrast to that, most robot-based operations such as pick-and-place motions or assembly tasks still use inefficient standard concepts causing high-energy consumption and high-energy costs. Thanks to a rather low payload-to-weight ratio of new robot designs, such as parallel kinematic or hybrid robot manipulators, a high potential for energy savings is expected. This article identifies potentials for energy saving concerning industrial consumers by assessing three different robot concepts. Based on a literature review, two existing designs for robots – the conventional serial robot and the parallel kinematic robot are analysed and compared with respect to the energy utilised during a typical item placement task. Afterwards, the concept of PARAGRIP, a hybrid of the two presented robot designs is introduced and examined based on simulation regarding its energy consumption. The final results demonstrate significantly different energy consumptions between the robot concepts, identifying potential savings of about 40% in a selected industrial application scenario.

Keywords:

• Industrial energy use
• serial robots
• parallel robots
• PARAGRIP

1. Introduction

In recent years, strong efforts have been made towards a sustainable utilisation of the world’s resources, exemplarily shown by electric mobility, renewable energies and a trend towards a carbon-neutral way of life. The annual growth of energy consumption in the industrial sector, however, still causes great concern among many countries. Accordingly, the question of energy efficiency became an important topic already in the 1970s, when the industry consumed more energy than any other end-use sector. From today’s perspective, the world energy consumption is estimated to increase by 1.9% per year in the period from 2010 to 2020. In this context, the industrial sector uses more energy than any other end-use sectors, consuming about 37% of the world’s total delivered energy in 2010 (McKane, Price, and de la Rue du Can 2008).

Furthermore, the industrial sector’s electricity consumption is projected to grow from 200 quadrillion Btu in 2010 to 307 quadrillion Btu in 2040, depicting an average of 1.4% each year between 2010 and 2040 (U.S. Energy Information Administration 2013 Figure 1). Moreover, the developing countries of Asia, Africa and South America will utilise great potential for economic growth and gain more and more shares of the world energy market (U.S. Energy Information Administration 2013).

Figure 1. Development of the world’s energy consumption.

Due to more international competition for energetic resources as well as a growing demand by successful industries, the price for electricity and other energetic resources will continue to rise. The rate of industrial energy consumption is important for the competitiveness of EU economies as it directly affects production costs. As pointed out by the European Union (European Commission 2014):

All future scenarios suggest there will be upward pressure on energy costs in the EU, not least because of the need to replace aging infrastructure, upward trends in fossil fuel prices, implementation of existing climate and energy policies and any impacts from a higher carbon price.

Typically, industrial approaches towards energy saving focus on the introduction of technologies with higher operative efficiency. Optimisation of input parameters, such as lead time, speed and temperature may additionally contribute to massive energy saving in industrial facilities. Other approaches are devoted to a better management of energy resources or to a reduction of energy leakage contributing to the overall energy-saving policy. In line with the implementation of energy-saving strategies, a considerable amount of emissions can be reduced (Abdelaziz, Saidur, and Mekhilef 2011). Other energy efficient methods that deliver significant outcomes vary from the introduction of renewable energy sources, optimal hardware selection and logistics planning, to reorganisation of the whole production plant. As stated by Pellicciari et al. (2013), the introduction of new technologies, management methods and energy policies has to be weighed against the economic and environmental impact: Besides economic aspects, there is also a socio-economic motivation to energy savings. According to Mahlia (2002), the industrial development across the world results in more energy use and leads to an increase in the emission of carbon dioxide (CO₂), sulphur dioxide (SO₂), nitrogen oxide (NO_x) and carbon monoxide (CO). These gases have disastrous environmental consequences for the earth’s climate, such as rising temperatures, droughts, floods as well as food shortages, causing economic chaos.

As stated above, the replacement of old equipment and hardware by more efficient ones becomes more and more important for the industrial sector. A lot of research has already been conducted in the area energy intensive processes connected to heat or cold as well as about pumps and auxiliary equipment with regard to more efficient alternative equipment. In several applications, selecting the most energy efficient concept for a specific task also proved to be the most economically feasible (see e.g. Duflou et al. 2012; Aurich, Carrella, and Steffes 2012; Kosanovic and Ambs 2000)

In contrast to that, industrial robot applications mainly rely on conventional serial robots, already existing since the 1950s. Nevertheless, selecting the right concept for a specific task is expected to contribute greatly to energy efficiency potentials. This research paper, accordingly, will address the following question:

Which contribution to energy efficiency can be gained from selecting the most suitable robot concept for industrial applications from an economic as well as an ecologic point of view?

The following article will thus research the main drivers for energy consumption for industrial robots originating from their basic design. In addition to the existing concepts of serial and parallel manipulators, the hybrid structure PARAGRIP will be introduced and included in the comparison. Besides an economic assessment, an ecologic assessment based on emissions of carbon dioxide will be provided.

2. Literature review

The following section provides a short introduction to the overall contribution of robots to the worldwide energy consumption in the industry. It will be shown that, even though robots are not a major energy consumer from an industry-wide point of view, they represent a decisive factor for specific applications, e.g. in assembly plants. Afterwards, three different robot topologies – the serial robot, the parallel kinematic robot and the hybrid structure PARAGRIP – are compared with regard to their basic design and resulting determining factors for their energy consumption.

2.1. Energy Consumption in the Industry by electrically powered Robots

According to the International Federation of Robotics IFR (2016), the worldwide operational stock of industrial robots increased considerably in 2013. The total worldwide stock of operational industrial robots at the end of 2013 was in the range of roughly 1.5 million units (see Figure 2). Over the last 10 years, this figure has annually grown by about 9.5%.

Figure 2. Annual world sales figures for industrial robots (IFR 2016).

Today, robots are major players in most of the manufacturing industries. Their application varies from single units in packaging tasks to more than a hundred in advanced assembly lines. Of course, the energy consumption of industrial robots depends on the specific application and thus varies case by case, nonetheless their growing number lead to a more significant energy consumption in the plants (Ystgaard et al. 2012).

Core processes in most industrial branches are predominantly operated by electrical motors. Concerning industrial robots, the main tasks consist of positioning operations and object manipulation, especially in assembly lines. As pointed out in Fleiter, Eichhammer, and Schleich (2011), recent studies discovered large energy efficiency potentials in electric motors with short payback times and high cost-effectiveness. In 2007, electric motor systems accounted for at least 60% of industrial electricity consumption in the USA (Saidur 2010). In order to estimate a global pattern of energy consumption by industrial robots, the indication given by the US figures is representative, as in 2007 according figures for other countries varied but showed a similar trend. Besides pumping, compressed air and fan systems, material handling and processing are some of the most electricity-consuming motor systems. The same trend has been observed by Siemens (2013), declaring that in industrial applications, about two-thirds of the total power consumption is accounted for by electric motors that are used to drive conveyor systems, machinery and industrial robots.

In a recent study, Javied et al. (2016) found that the overall energy consumption of robots in the German industry can be accounted for about 1% of the total electrical energy consumption. This seems to be a rather small amount in comparison to e.g. 12% for machine tools or 6% for rolling and pressing of metals. Still, this study considered the entire German industrial sector. For certain companies and industry sectors, the energy consumption of robots can make up for a much larger part of the total: e.g. Müller and Löffler (2009) presented a deep analysis of the energy consumption of a car body shop (see Figure 3).

Figure 3. Exemplary energy consumption of a car factory according to Müller and Löffler (2009).

Its profile represents a mix of energy sources, where electricity is the dominating one with costs reaching 82% of total energy costs at the plant. In the same car body shop, the energy consumption per equipment is researched. During the manufacturing phase, robots are identified to be the second-largest energy consumers, accounting for 20% of the main peripherals. In addition to that, they appeared to be the second-biggest energy wasters during standby time.

In recent years, research activities on the energy consumption of industrial robots have been focusing on two main aspects: On the one hand, the physical robot cell can be optimised i.e. by an efficient cell configuration and selection of the kinematic topology and structure, by an efficient mass distribution and lightweight design of the robot or by an economically optimised dimensioning of the actuator concept and control unit. On the other hand, the robot application can be improved, i.e. by an optimised task management and motion design or by energy recuperation. Research on efficient motion design for stationary, industrial robots started with early contributions by Neuhaus(1990) and Park (1996), who developed minimum energy velocity profiles for one-dimensional trajectories. Subsequent approaches primarily analysed the energy efficiency of multidimensional trajectories. Accordingly, Delingette, Hebert, and Ikeuchi (1991) considered the optimisation of spline functions with minimum energy curvature constraints. Further approaches established promising elliptical (Katoh et al. 1994) or sinusoidal trajectories (Diken 1994), which, however, were limited to special applications. In spite of numerous research approaches towards a sustainable and efficient task and motion planning, a detailed comparison of possible robot topologies has not been considered yet. Accordingly, this contribution is devoted to the energetic classification of different robot topologies, whose energy consumption is compared by the evaluation of typical motion tasks. In the following paragraphs, two major designs will be examined with respect to their energy consumption – the conventional serial robot and the parallel kinematic manipulator (see Figure 4(a) and (b). Afterwards, the manipulator PARAGRIP will be introduced, representing a hybrid topology, which combines numerous advantages of both designs (see Figure 4(c)).

Figure 4. Different robot topologies. (a) serial topology, (b) parallel topology, (c) hybrid topology.

All three concepts in Figure 4 feature very diverse designs resulting in differing physical properties and mass distributions and thus also differing energy consumption per motion. These will be reviewed briefly in the next subchapters.

2.2. Serial industrial Robots

The energy consumption of a serial industrial robot depends on the equipment which is needed for the respective application. Independent of the task, robots need energy to overcome friction occurring in the joints and to be able to move or hold loads (Ystgaard et al. 2012). Serial robots consist of relatively heavy components, which only can only be moved with high-energy expenditure. There are two reasons for their own high weight. The first aspect is given by the robot’s high structural stiffness, which is achieved via a rigid and solid design and by heavy materials for each link. Accordingly, the high stiffness is supposed to ensure a high positioning accuracy. The other reason is given by the serial structure itself providing an alternating sequence of actuated joints and links. Consequently, one servomotor is required for each joint, additionally leading to the high weight of serial robots (Shaik, Tlale, and Bright 2013).

Due to the design of serial industrial robots, they permanently consume energy in order to carry their own weight. The standby mode makes up a high proportion of the total energy consumption (International Federation of Robotics IFR 2016). According to Ystgaard et al. (2012), this sums up to even the biggest amount of the entire energy consumption. Furthermore, the place of installation, the trajectory planning and the used workspace have to be considered with respect to energy consumption. Large horizontal distances between the robot base and the manipulated object’s centre of gravity, for example, induce high actuator torques. In vertical direction, additional energy is necessary to overcome gravitational effects leading to a higher energy consumption in this direction (Chemnitz, Schreck, and Krüger 2011).

Figure 5 shows the electrical power performance of a vertically articulated serial robot for carrying an object in two different positions (Ystgaard et al. 2012). As soon as the control system boots, a basic energy supply for the standby mode is required. Additionally, the load can be carried with or without motor brake. If the motor brake is turned on, the power requirement is independent of the position on the horizontal x-axis. In Contrast, as soon as the brakes are turned off, the required power increases depending on the distance between end-effector and robot base due to a higher torque. The graph illustrates that the maximum power requirements result from respective single motions causing power peaks. However, when observed over the entire runtime, power peaks are not responsible for the main energy consumption as the standby mode and the task of keeping the robot arms in certain positions consume much more energy. As a consequence, the operating time has to be as short as possible (Chemnitz, Schreck, and Krüger 2011). If this shall be achieved by increasing the speed of motion, it has to be considered that more energy is needed to accelerate and decelerate the robot arm again. The required energy for braking, however, can be reduced by an intelligent trajectory planning. Avoiding abrupt movements reduces the acceleration of the robot arm and thereby its energy consumption.

Figure 5. Energy consumption for different movements of a serial robot as presented in (Ystgaard et al. 2012).

An alternative approach to save energy is the use of programmable power-down systems. These switch off or reduce the power of electronic components according to their minimum necessary consumption. In particular, this offers a high potential of energy savings during idle phases (Chemnitz, Schreck, and Krüger 2011).

The existing energy consumption optimisation methods schedule the performance of industrial robots at their maximum operational speed and assume that the robot is standing still the rest of the available time. Recent research by Pellicciari et al. (2013) points out that the described approach may be energetically inefficient due to high-power consumption peaks and longer waiting times, while energy is wasted to compensate the gravitational loads. As stated by Meike and Ribickis (2011), the idle and homing times for industrial robots in the automotive industry vary from 15 to 20% of the total working times in some cases and up to 74% in other cases.

According to Ystgaard et al. (2012), the main findings for the industrial application of serial robots with regard to their energy efficiency are given by the relation between energy consumption, velocity and operational times: Higher velocities and accelerations lead to a lower total energy consumption, while the mass of the payload has only a small influence on total consumption. In addition to that, the standby consumption is a dominating factor for energy efficiency.

2.3. Parallel kinematic manipulators

A parallel kinematic manipulator is a robot system consisting of a manipulated mobile platform, also called end-effector, which is connected to a base frame by at least two actuated kinematic chains. A simplistic example is presented in Figure 6. While serial robots represent one open kinematic chain, parallel kinematic manipulators use several chains in parallel, enabling the positioning of actuators in the base frame. Due to this design, parallel manipulators feature a much smaller workspace in comparison to their footprint. As pointed out by Mbarek, Nefzi, and Corves (2005), the lightweight design, however, provides a much better performance of parallel manipulators in several other parameters such as a high payload-to-weight ratio and better dynamics as well as a higher accuracy and stiffness. Especially, the small self-weight and the high payload-to-weight ratio result in a comparatively low-energy consumption.

Figure 6. A simplified structure of a parallel manipulator (Mbarek, Nefzi, and Corves 2005).

In a contribution by Li and Bone (2001), the average energy usage of parallel kinematic manipulators was determined to be 26% of the one given by serial manipulators, considering the same drive motors and a similar workspace. At a first glance, this is not surprising due to the much smaller weight of the moving parts of the parallel manipulator compared to a serial robot. The authors, furthermore, pointed out, that the high efficiency was only slightly affected by end-effector velocity, acceleration or static loading. However, Thanh et al. (2012) noticed the significant disadvantages of parallel robots in contrast to serial ones, such as a small workspace compared to the occupied space, the singular configurations within the workspace and challenges in dynamics modelling due to multiple closed-loop kinematic chains.

According to Li and Bone (2001), parallel manipulators are less space efficient and thus have to be at least twice as large as a serial manipulator to cover the same workspace. Even though these robots outperform serial ones with respect to energy efficiency, they might not be applicable for a large amount of industrial applications.

2.4. The hybrid structure PARAGRIP

PARAGRIP is a self-reconfigurable three- or four-arm robot, designed for low-energy and cost-efficient material handling and processing operations of large scale components. An illustration is given in Figure 7. The system offers some of the advantages of parallel manipulators such as flexibility, high operational speed as well as lightweight and mobile arms while performing the work of a whole system of standard industrial robots (Mannheim et al. 2013). Due to a novel structure, it overcomes the technical performance of a group of industrial robots working together. According to Corves et al. (2012), the robot structure can be adapted with respect to a specific task and self-optimised to suit a selected feasible layout, regrasp point, base configuration or regrasping operations by the use of reconfiguration.

Figure 7. The kinematic structure of the PARAGRIP system.

The combination of serial and parallel kinematic structures leads to a bigger workspace than classical parallel structures and also induces a better payload-weight ratio compared to classic serial structures. The system furthermore offers advantages with regard to versatility and reconfigurability. The possibility of reconfiguration and self-optimisation offers a fast and easy adaption of kinematic parameters based on the requirements of changing tasks, influencing the main motion characteristics e.g. workspace, stiffness and accuracy. Handling objects of different shape and size can be manipulated without a specialised gripper. PARAGRIP will be used for the assembly of large-scale components as part of the research project ‘Integrative Production Technology for High Wage Countries’ (Corves et al. 2012).

The system outperforms parallel manipulators with regard to flexibility and workspace and could thus be applied to a broader field of industrial scenarios. Due to its basic design, it is expected to consume far less electrical energy than a serial robot due to the small inertia and improved payload to weight ratio. As no field data are available yet, we offer a sophisticated simulation of the energy consumption of PARAGRIP in comparison to a serial robot in the following chapter to get a first quantitative estimation for potential energy savings (Table 1).

Table 1. Qualitative comparison of the characteristics of the different robot concepts.

Characteristic	Serial robot	Parallel robot	Paragrip
Workspace/installation space ratio	Good	Poor	Mean
Payload/weight ratio	Poor	Good	Mean
Dynamic potential	Mean	High	Mean
Required drive torque	High	Low	Low

2.5. Interim conclusion

It can be concluded that a lot of scientific and technical effort has been invested into optimising single robotics technologies with regard to energy efficiency. Most work has been focused on serial robots, as they make up the largest share of the total units in use. Depending on the design of a robot concept, basic assumptions regarding its energy efficiency in contrast to other designs can already be made with regard to the motion of masses and inertia, payload-to-weight ratio etc. As expected, parallel manipulators can perform tasks with less energy per motion, when compared to serial robots. On the other hand, they can only be operated within a limited workspace. In contrast to that, the manipulator PARAGRIP depicts a compromise between these two concepts as it features a larger workspace and a higher payload-to-weight ratio leading to a higher expected energy efficiency. In the scientific literature, a direct comparison of the PARAGRIP with e.g. serial robots has not been made yet with respect to its energy consumption and its impact on economic and ecologic life cycle. According analyses will be presented in the next chapter.

3. Economic and ecologic impact of PARAGRIP – a simulative comparison

To assess the economic and ecologic impact of a technical system, the ISO 14044 standard has become one of the world’s most used approaches (Benoît et al. 2009). Its basic structure consists of four steps (Klöpffer and Renner 2008), which are depicted in Figure 8: In the beginning, the goal and scope of the analysis needs to be derived. This includes the objective of the overall comparison, the system boundaries, location, time frame and major components for the assessment. In a second step, the inventory analysis is provided: every in- and output variable of the system is registered. Afterwards, the impact of these variables needs to be assessed. Depending on the goal and scope, the impact can vary significantly: e.g. the cost of the usage of one kWh of electricity depends on the exact geographical location of the system, as well as the contracts with the local energy supplier. The ecological impact of its usage varies in the same way, as e.g. German electricity production is heavily based on fossil fuels, while Scandinavian countries use mostly renewable sources. Thus, the analyses of the inventory and the interpretation of the impact need to be separated. As a last step, the interpretation has to be executed, putting the findings of the impact analysis into context. Again, depending on the economic and ecologic environment and situation, the impact of the same object or process can vary significantly (Finkbeiner 2011).

Figure 8. The four major steps of the ISO 14044.

In the context of this contribution, a first analysis based on the ISO standard will be executed in order to compare the PARAGRIP system with a serial robot with regard to its energy consumption for specific payload motions to assess its impact on the economic as well as ecologic life cycle (goal and scope). In order to provide a wide range of test data, the inventory analysis will be based on a simulation model of the two systems. The impact assessment will be carried out based on ecologic factors such as emission-equivalents of CO₂ and other available material, based on a usage in western Europe, e.g. Germany (due to availability of electricity prices as well as emission-equivalency factors). The interpretation will be done in the context of a production as can be found in western European countries.

3.1. Design of the simulation model

To compare the energy consumption in different scenarios, PARAGRIP as well as two different sized serial robots have been modelled in the Multibody-System Matlab/SimMechanics. The two different robots were chosen by reason of non-comparability of workspace and payload. As the payload of the PARAGRIP system is 10 kg, a medium-sized robot has been chosen, which is able to handle the same payload. Hence the workspace of a 10 kg payload robot is much larger than for the PARAGRIP system and the system mass is much higher, thus the energy consumption is expected to be worse. For these reasons, a small serial robot with comparable workspace has been introduced to the simulation additionally. This robot features a lower payload of about 5 kg. Hence, a reduced handling task has been defined in addition to the first one.

As this is a brief estimation of the required energy, it is assumed that the actuators of all robots have the same energy conversion efficiencies. This means the drive efficiency factors are just a constant multiplication and can be neglected in this comparison. The energy consumption is measured by the integral of the power needed for each time step. The power is computed by multiplying the actuators angular velocity ω_i with their torque T_i, where is the index of the considered drive:

The corresponding equation makes use of the angular velocity ω_i as well as the drive torque T_i in joint space, whose vectorial representation can be determined using the robot’s Jacobian matrix . Accordingly, the vector of actuator velocities results from multiplying the Jacobian with representing the end-effector velocity:

The vector of drive torques results from multiplying the transposed inverse of the Jacobian with the vector of external forces :

In this study the drives are not able to recuperate, otherwise the utilised drives would have to be measured in order to get a reliable result on energy recuperation, which is not practicable within this first estimation. The modelled robots are shown in Figure 9.

Figure 9. Models of the two robot concepts as used in the MATLAB simulation.

As the robot’s energy usage is highly depended on the end-effector position and the trajectory in the workspace, random generated paths covering the volume of a desired cubic workspace are used. This preserves in choosing trajectories, which are best for one of the robots. The curves are generated by Lissajous-curves (Riedel 2014), as they are periodic functions and are, thereby, infinitely differentiable:

Due to this, the curves have moderate normal acceleration profiles without discontinuities. Accordingly, the Lissajous-curves represent an adequate method to generate paths for this kind of comparison, as there are no artificial drive torque peaks. An example with 10 curves is depicted in Figure 10.

Figure 10. Exemplary randomised Lissajous-curves for the evaluation.

3.2. Execution of the simulation

The described paths are the same for each robot and limited to a cube with 0.4-m lateral length to ensure the physical restrictions of the different systems. Hundred curves are generated to obtain a representative set, but still having an acceptable computing time. The overall motion time of each path is set to be 1 s, giving a maximum velocity of 2 m/s. For this study, the specialised robotic system PARAGRIP has been compared to standard industrial robots available on the market. Thus, two Kuka standard robots were chosen. Taking the maximum payload into account, the Kuka KR 16-2 with a maximum payload of 16 kg is the nearest match. Additionally, the Kuka KR 6 R900 six is chosen by the size of the workspace as small-sized robot with reduced payload of 6 kg. Thus, the simulated task is defined by a load of 5 kg to compare all three robots and 10 kg to compare the bigger one with PARAGRIP.

3.3. Major findings of the simulation

The theoretical simulations allow a qualitative comparison of the consumed energy for the robotic tasks. As the validation of the models with the real robots has not been done yet, the simulation just shows the energy consumptions order of magnitude of the different robotic concepts. The results show that the midsize robot requires a lot more energy than the other ones. This fact can be explained by the size of the workspace, and by the high mass of the robot which has to be moved. The task requires only a small volume of its theoretical workspace. Therefore, the midsize robot needs most of the energy to move itself. When looking at the task of moving a mass of 5 kg, the medium-sized robot requires more than six times more energy than the PARAGRIP system, whereas the small industrial robot shows better performance but still requires almost double the energy.

Based on first measurements at the prototypical build-up, it can be estimated that the PARAGRIP system should perform better. Still, a more exact examination is needed and will be the subject of future research. Comparing just the results of the industrial robots, they already show how much energy is wasted by choosing the wrong-sized robot.

3.4. Economic and environmental Impact and interpretation

The simulation results clearly show that the choice of robot concept and size has a large influence on the energy consumed for a specific task. Even if it might not always be feasible to look for alternative specialised concepts such as PARAGRIP, even choosing a modest size unit and applying methods such as shut down in idle times offer a great potential for energy efficiency.

The PARAGRIP manipulator offered a savings potential between 40 and 80% in energy consumption when compared to different standard industrial robot concepts. In the application example of Ystgaard et al. (2012), a total energy of about 3,889 Wh was consumed in 30 s. Repeating this task continuously would lead to a consumption of 466 Wh per hour of operation. This might not seem a lot, but robots very often repeat the same task for several 10.000 h of service in mass producing facilities. So, based on the prices pointed out by the European Commission (2014), a reduction of 40 to 80% in electricity consumption would lead to savings between 210 € and 430 € per 10.000 h of service. As the prices for electricity and energy resources are still on the rise, this potential will grow as well.

In addition to the reduction of cost, large ecological saving potentials can be obtained from choosing the right-sized and most suitable robot concept. Table 2 provides a short overview of the resulting ecological impact of using the three different robots for the task of moving 5 or 10 kg as depicted in the previous chapter. In this calculation, CO₂-equivalents (which can be found in [Covenant of Mayors for Climate & Energy 2016]) as well as radiation U235-equivalents (as depicted by Thinkstep Global Inc. 2016) have been used. As can be deducted, having the same robot run in Germany causes a much higher CO₂-emission than in France, due to the completely different energy mix. While the German energy supply heavily relies on coal and gas, France largely operates nuclear power plants, causing other emissions with long-term environmental impact (Table 3).

Table 2. Comparison of mean energy consumption for different robots and different payload based on the Lissajou-figures in Figure 10.

Payload/energy consumption (kg)	PARAGRIP (kJ)	KR 6 R 900 six (kJ)	Improvement (%)	KR 16 (kJ)	Improvement (%)
5	7.87	14.13	44	47.63	83
10	11.4	–	–	52.1	78

Table 3. Environmental impact of the different robot concepts.

	PARAGRIP	KR 6 R 900 six	KR 16
Payload 5 kg
Energy (kJ)	7.87	14.13	47.63
CO2 equivalent (Germany) (g)	1.36	2.45	8.26
CO2 equivalent (France) (g)	0.12	0.22	0.74
Radiation: U235 equivalent (Germany) (g)	0.59	1.06	3.57
Radiation: U235 equivalent (France) (g)	2.43	4.36	14.69
Payload 10 kg
Energy (kJ)	11.4	–	52.1
CO2 equivalent (Germany) (g)	1.98	–	9.03
CO2 equivalent (France) (g)	0.18	–	0.81
Radiation: U235 equivalent (Germany) (g)	0.86	–	3.91
Radiation: U235 equivalent (France) (g)	3.52	–	16.06

This means that running the large KUKA KR 16 for 10.000 operating hours steadily repeating the task of moving 10 kg back and forth will cause an emission of 3.25 t of CO₂, while the emission cause by same task on the PARAGRIP will be about 700 kg. It can be deducted that, even though industrial robots tend to have a minor effect on electricity consumption on an industry- or even nation-wide level, the span between different concepts and sizes will be relevant. Selecting the right-sized and most suitable concept for a specific task shows large impacts on its economic and ecologic lifecycle.

4. Outlook and future work

The simulation presented above provided an estimation of the potentials for energy savings of differing robot concepts and sizes, especially the hybrid structure PARAGRIP. In the course of the Cluster of Excellence ‘Integrative Production Technology for High Wage Countries’, this robot concept in particular will be subject to further research regarding its potentials and integration into a real industrial assembly task. Building up the system will finally provide the ability to validate the simulation results stated above. Further research topics will be in the area of grasping technology, human–machine interaction and more intelligent production control.

5. Summary

With rising cost for energy and its related resources as well as a growing political pressure, the increase in energy efficiency has become more and more important for industrial consumers. Energy-intensive processes such as heating, cooling or material transformation have been in the focus of scientific research as well as applied optimisation projects. In contrast to that, the energy consumption of industrial robots received only minor attention due to the rather small overall impact.

As the payload-to-weight ratio of standard industrial robots appears to be rather small, energy efficiency potentials can be found in its optimisation as well as the use of different robot concepts. This contribution reviewed the energy consumption of standard industrial units in comparison to parallel kinematic ones. In a second step, the PARAGRIP system as a hybrid approach between serial robots and parallel kinematic manipulators has been introduced. Its energy consumption has been determined in the way of a computer simulation and proved to be between 40 and 80% better in contrast to standard industrial robots.

The right selection of robot systems with respect to their kinematic structure as well as their size yield large energy efficiency potentials per payload motion. For a single motion or task, the potential seems to be rather small. However, with energy costs on the rise as well as a growing political pressure and when looking at the large number of units in service word wide, these energy efficiency potentials will become more and more important in an industrial context.

Disclosure statement

No potential conflict of interest was reported by the authors.

Funding

This work was supported by Deutsche Forschungsgemeinschaft [grant number EXC-128].

Notes on contributors

J. Kurilova-Palisaitiene works as a research fellow of the Department of Management and Engineering, Division of Manufacturing Engineering at Linköping University. Her main interests in research are lean management, sustainability and product recycling.

E. Permin, Dr.-Ing., is working at the Fraunhofer Institute for Production Technology as Head of the Department of Production Quality. He is also a lecturer at the RWTH in Aachen where he teaches Quality Management. The focus of his scientific work is in the area of artificial intelligence for production systems and the usage of smart devices on the shopfloor.

T. Mannheim, Dr.-Ing., worked as a scientific employee at the Department of Mechanism Theory and Dynamics of Machines, RWTH Aachen University, Germany. His main research interests are in software development for mechanism design, energy efficiency potentials of robot concepts and motion control of robot structures. Previous publications have appeared in Applied Mechanisms and Materials, Journal of Robotics and Computer-Integrated Manufacturing, Mechanisms and Machine Science, and others.

K. Buhse worked as a student researcher at the Department of Mechanism Theory and Dynamics of Machines, RWTH Aachen University, Germany. His main research interests are in energy efficiency potentials of robot concepts and design of weaving machines. Previous publications have appeared in Applied Mechanics and Materials.

M. Lorenz works as a scientific employee at the Department of Mechanism Theory and Dynamics of Machines, RWTH Aachen University, Germany. His main research interests are in parallel kinematic manipulators, robot redundancy, efficient trajectory planning of robot manipulators and human-robot-collaboration. Previous publications have appeared in New Advances in Mechanisms, Mechanical Transmissions and Robotics, Journal of Robotics and Computer-Integrated Manufacturing, Proceedings of the IEEE International Conference on Robotics and Automation, and others.

R. Schmitt, professor Dr.-Ing., works with the Institute of Metrology and Quality Management, Laboratory for Machine Tools and Production Engineering (WZL), RWTH Aachen University, 52074 Aachen, Germany. He is the member of the board of the Fraunhofer Institute for Production Technology IPT as well as of the board of the Laboratory for Machine Tools and Production Engineering. He is member Institute of Electrical and Electronics Engineers (IEEE), German Society for Quality e.V. (DGQ) and Society for Quality Science e.V. (GQW).

B. Corves, Univ.-Prof. Dr.-Ing. Dr. h. c., is the director of the Department of Mechanism Theory and Dynamics of Machines, RWTH Aachen University, Germany. He has more than 25 years of research experience in the fields of mechanism theory, robotics and dynamics of machines. Previous publications have appeared in journals (e.g. Journal of Robotics and Computer-Integrated Manufacturing) and conference proceedings (e.g. Applied Mechanisms and Materials, New Advances in Mechanisms, Mechanical Transmissions and Robotics) in various areas such as mechanism theory, robotics, bio kinematics, machine dynamics and vibrations.

M. Björkman, professor PhD, is head of Manufacturing Engineering in the Department of Management and Engineering, Division of Manufacturing Engineering, at Linköping University. He is a senior researcher in the area of production systems, supply chain management and product design with a keen interest in lean management, sustainability, lifecycle assessment and biotechnology.

Acknowledgements

The authors would like to thank the German Research Foundation DFG for the kind support within the Cluster of Excellence “Integrative Production Technology for High-Wage Countries”.

Appendix 1.

Used masses

PARAGRIP-Prototype (each arm)

Table

Link	Length (m)	CG (m)	Mass (kg)	Mass (%)	Inertia
Full	_____	_____	6.976	100	_____
1	L1 = 0.278	[0.142,0, 0]	3.1	44.4	[52.952e-3 0 0
					0 50.962e-3 0
					0 0 4.364e-3]
2	L2 = 0.68	[0.387,0,0]	1.68	24.1	[2.152e-3 0 0
					0 30.888e-3 0
					0 0 30.737e-3]
3	L3 = 0.48	[0.0023,0,0]	0.74	10.6	[0.695e-3 0 0
					0 20.972e-3 0
					0 0 20.904e-3]
4	L4 = 0.19	[0.132,0,0]	0.136	1.9	[12e-6 0 0
					0 2.3e-3 0
					0 0 2.3e-3]
5	L5 = 0	[0,0,0]	0.97	13.9	[0.836e-6 0 0
					0 2.221e-3 0
					0 0 2.408e-3]
WJ	L6 = 0.158	[0.05,0,0]	0.35	5	[0 0 0
					0 0 0
					0 0 0]
Kuka KR 16-2
Full	_____	_____	235	100	____
0	L0 = 0.391	_____	58.24	36.4	____
1	L1y = 0.278	[0.142,0, 0.1770]	42.88	26.8	[0.5719 0 0
	L1x = 0.260
					0.5719 0 0
					0 0 0.3326]
2	L2 = 0.68	[0.387,0,0]	22.4	14	[0.0857 0 0
					0 1.3106 0
					0 0 1.3106]
3	L3 = 0.48	[0.0023,0,0]	27.04	16.9	[0.1171 00
					0 0.8211 0
					0 0 0.8211]
4	L4 = 0.19	[0.132,0,0]	6.24	3.9	[0.0181 0 0
					0 0.0366 0
					0 0 0.0366]
5	L5 = 0	[0,0,0]	0	0	[0 0 0
					0 0 0
					0 0 0]
6	L6 = 0.158	[0.05,0,0]	3.04	1.9	[0.0082 0 0
					0 0.0134 0
					0 0 .0134]
KR 6 R900 sixx
Full	_____	_____	52	100	____
0	L0 = 0.217	_____	13	25	____
1	L1y = 0.183	[0.008,0,0.116]	11.284	21.7	[0.0369 0 0
	L1x = 0.025
					0 0.036 0
					0 0 0.0109]
2	L2 = 0.455	[0.203,0,0]	16.38	31.5	[0.0342 0 0
					0 0.2997 0
					0 0 0.2997]
3	L3 = 0.107	[0.016,0,0]	5.1428	9.9	[0.0084 0 0
					0 9.1e-3 0
					0 0 9.1e-3]
4	L4 = 0.317	[0.147,0,0]	5.408	10.4	[0.0113 0 0
					0 0.0509 0
					0 0 0.0509]
5	L5 = 0	[0,0,0]	0	0	[0 0 0
					0 0 0
					0 0 0]
6	L6 = 0.073;	[0.0130,0,0]	0.8164	1.6	[0. 506e-3 0 0
					0 0.615e-3 0
					0 0 0.6157e-3]