A Common Software Framework for Energy Data ... - Science Direct

Available online at www.sciencedirect.com

ScienceDirect Procedia CIRP 61 (2017) 359 – 364

The 24th CIRP Conference on Life Cycle Engineering

A Common Software Framework for Energy Data Based Monitoring and Controlling for Machine Power Peak Reduction and Workpiece Quality Improvements Christoph J. H. Bauerdicka*, Mark Helferta, Benjamin Menzb, Eberhard Abelea * a

Institute of Production Management, Technology and Machine Tools (PTW), Otto-Berndt-Str. 2, 64287 Darmstadt, Germany b Bosch Rexroth AG, Rexrothstr. 3, 97816 Lohr am Main, Germany

* Corresponding author. Tel.: +49-6151-16-20128; fax: +49-6151-16-20087. E-mail address: [email protected]

Abstract A standardized and high frequent collection of energy data is a powerful tool to enhance industrial energy efficiency and has also proved to be most valuable in condition and process monitoring tasks in recent researches. For these monitoring and controlling functions in production sites a common software framework is essential. In this paper the development of such a software framework is presented. Further on two new approaches utilizing this software framework are introduced: Firstly the recognition and automatic reduction of power peaks between connected machine tools. Secondly the usage of energy data of machine tools for quality monitoring purposes. © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license © 2017 The Authors. Published by Elsevier B.V. (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the cscientific of theConference 24th CIRP on Life Cycle Engineering. under responsibility of the scientifi committeecommittee of the 24th CIRP onConference Life Cycle Engineering Peer-review Keywords: Software Framework; Production Machines; Energy Controlling; Industry 4.0; Quality Monitoring; Condition Monitoring; Workpiece Monitoring

1. Introduction 2014 was the warmest year on record. Compared to the global average of the 20th century the temperature was 0.69 °C above [1]. The greenhouse gas emissions, which are one main factor of these global ecological trends, of 49.5 GtCO2 eq/yr are mainly driven by the industry with 28.6 % in 2010 [2]. The industry in Germany, which uses 30 % [3] of primary energy in 2011 and 46 % of electrical energy in 2013, must optimize the energy consumption with different strategies to stay competitive as well as to ensure that the ecological pollution stays in an acceptable range for humans, biodiversity and ecological systems. There is not just the necessity to increase the efficiency in industrial processes and its components which leads to less energy and resource consumption: Renewable energy sources like wind energy or solar energy are highly volatile. Electric energy usage must be able to follow up a given profile or high load peaks must be prevented.

In the research project ETA-Factory challenges in energy efficiency for production machines, factory buildings and technical infrastructure are addressed [4]. For example a thermal network, which is established, can increase the energy efficiency using waste heat between machines as well as between machines and the building itself [5]. Individual optimization of subsystems (machines or building) reaches 30 % energy efficiency rates, whereas overall optimization leads to 40 % as presented in [4]. A mayor part of the energy savings are based on the connection of machines and devices. According to [6] connected devices in the different fields of home consumer, transport and mobility, health and body, buildings and infrastructure, cities and industry will grow from 2 billion to 8 billion devices in 2020. This increase will change the market volume from 180 billion to 1 trillion $/yr and with the growth of connected devices in any domain, the opportunities of solving technical problems will grow. In its hightech strategy [7] the German government focusses on digitalization and energy efficiency in the industry. It is

2212-8271 © 2017 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Peer-review under responsibility of the scientific committee of the 24th CIRP Conference on Life Cycle Engineering doi:10.1016/j.procir.2016.11.226

360

Christoph J.H. Bauerdick et al. / Procedia CIRP 61 (2017) 359 – 364

mentioned that Industry 4.0 is one enabler for linked energy systems to increase energy efficiency as shown in [8]. Connected devices are IT infrastructural linked objects like PLCs or sensors. Together with a common software framework they enable energy optimization potentials. These actual trends of connected devices create a chance to optimize the energy consumption establishing a common software framework. In [9] a concept for analysing energy data is presented with low cost energy sensors. There is a high potential analysing energy data for recognizing energy waste in a factory as well as identifying bad states of processes. Transparency of energy consumption is one key enabler for energy optimization [9]. Especially clustering the information for different operational periods increases the motivation for energy efficiency topics [10]. The software framework presented by [11] is of a conceptual high level infrastructure of information systems for energy based management. In this work a method, especially for connected machine tools for data acquisition and analysis is presented. Focusing on the different levels on energy usage like micro (data in microseconds) and macro (data clustered hours) presented in [12], here the micro level is focused. Subsequently two approaches for the utilization of energy and process parameters based on the common software framework are demonstrated: power peak shifting and quality monitoring of workpieces. Power peak shifting for short term is effecting costs for infrastructure as well as a more balanced electrical grid. The concept can provide: x Optimization of energy consumption in a defined time interval and x Online power peak reduction for connected machine tools. The prospects of energy data (e.g. power demand) for quality monitoring during machining are analysed using simple signal processing methods based on an experimental series. It is shown that energy data is not only valuable for energy efficiency improvements but, providing the necessary software framework, for different condition monitoring purposes as well. 2. Software framework for energy monitoring and controlling of IT-connected machine tools

2.1. Framework and goals of GDS The Generic Data Server (GDS) software framework described in [14] was a system for analyzing data and restricted to special cases. For use cases, which are described in this paper, it is further developed to a workflow based structure for enabling more flexibility. Now analysing, monitoring and controlling machines using the communication protocol OPC UA is possible. Besides the workflow the GDS consists of an analysis section which is a Web based HMI as well as a database system for long term data acquisition (Fig. 1).

Fig. 1. Generic Data Server overview

Workflows are created through a configuration interface. Thereby a high flexibility is given to run tasks like x x x x

reading data, processing data, storing data and writing commands to any OPC UA capable PLC.

A simple workflow shown in Fig. 2 has a periodic trigger which activates the workflow “Processing”. With the following read activity data from an OPC UA server is read (like a temperature value or an energy counter from a PLC). User defined processing is possible using the “ProcessMatlab” activity which can run any Matlab created DLL. The data is transferred between each activity inside of the software with variables. Finally the processed data is stored in a database with a write activity.

In the following a software framework is presented which is improved for automatic data aggregation to be used in a workflow system. Internal sensors of e.g. drives are used for data acquisition. Therewith no additional sensors are needed for monitoring and controlling purposes. The standard for data exchange is OPC UA due to its security, stability, scalability and supplier independency [13]. Fig. 2. Workflow structure of the Generic Data Server [15]

Limitations in the rate of communication are actually given through OPC UA with 10 Hz. Especially in the area of power

361


peak shaving and quality monitoring, higher rates are necessary. 2.2. Framework and goals of EWB Initially the Bosch Rexroth Efficiency Workbench (EWB) [16] was intended to be used in a very static way to analyse process and energy data of one machine according to its physical production process (NC-Code). Therewith individual optimizations of cycle time and energy optimizations were possible. For continuously, NC correlated and high frequent data the EWB can now be used in the GDS. Data from internal drive sensors is accessed and recorded. In [14] a file streaming mechanism of EWB is presented. This is improved by streaming now data directly to a database of the GDS (Fig. 3). Fig. 5. EWB with Post Processing at the GDS system

The EWB can record data with a rate of up to 1 kHz. The power peak shaving and quality monitoring for machine tool is now possible. Fig. 6 shows the power demand of an EMAG VLC100Y with power peaks typically being around 50 ms (20 Hz). 25

Aggregation Spindle and Axes Hydraulic Pump Cooling System

Power [kW]

20

Fig. 3. EWB with Post Processing at a Machine PC

15 10 5 0 0

100

200

-5

Thereby the implementation into a workflow using a TableChangeTrigger is realized, as well as a lean and common data storage structure. Each instance of EWB (Fig. 4) can now stream data vectors into the central GDS using OPC UA communication. A connected machine tool data acquisition, which means data acquiring from IT-connected machine tools, is possible.

300

400

500

600

Time [s]

-10

Fig. 6. Power demand of peripheral devices and axis of an EMAG VLC 100Y (turning machine)

3. Power peak reduction 3.1. Power peak reduction concept for connected machines A typical start up peak of one machine is shown in Fig. 7. There is a height of 55 kW, where the continuous producing procedure ranges between 7 kW and 12 kW. This high peak is caused by the simultaneous start of

Fig. 4. Central collection of data with the GDS using EWB

Data post processing, which is the conversion of binary data, is moved from machine PCs (Fig. 3) to a workflow (Fig. 5). Thus no additional CPU capacity on a different System is necessary.

x spindles acceleration, x axis and x peripheral devices like hydraulic, cooling system, pressure air and PLC. The peak inside of one machine is already reduced by equipment like compensator plates. Many machines in one electric grid need a technical infrastructure according to their

362


electrical conditions to prevent overloads by peak load appearing.

Fig. 9. Data acquiring, analysing and reacting for peak preventing by shifting production processes

Fig. 7. Typical start up peak of a machine tool

Synchronizing machines in one electric grid would reduce that high impact on infrastructure as well as on the grid itself. The concept presented in Fig. 8 is used to shift power peaks occurring on different machines at the same time with

Data is acquired through EWB or PLC libraries and analysed in the GDS using Matlab built DLL. Results from the Matlab script are used to control the production process. This can be shifting times for a machine as well as pulsing for example a heating oven. The GDS writes into a PLC variable which is available in the NC Code. 3.2. Power peak reduction of connected machines Running the workflow with two connected machine tools a peak reduction is realized (Fig. 10). Initially a peak of 32.7 kW is reduced by 12.8 kW to 19.9 kW, shifting the production process of two machines automatically against each other.

x changing production order, x shifting production start and x start peripheral devices at power depth.

35

Standard

(b)

Optimized

Power

Power

(a)

Time Po we r

Machine 1

Time

Standard Machine 2 Ti

Aggregation

Fig. 8. Conceptional power peak of 2 machines at the same time (a) Standard; (b) shifted peak of one machine for less grid load

For shifting production processes the NC Code can be influenced. Accessing the NC Code from GDS is realized using variables in a PLC as shown in Fig. 9.

Power sum of two machines Optimized power sum of two machines

30 25 20

Power [kW]

In this case Fig. 8 shows peaks of two machines before and after optimization. Occurring at the same time causes high loads in a grid. Now shifting the process of machine 2 reduces the peak and ensures a smaller load in the grid. The typical load cycle is automatically measured for each connected machine tool, using EWB. Therefore one machine has to wait for one production cycle, which means very low productivity losses in highly automated production areas with high production quantities.

15

10 5 0 150

170

190

210

-5

230

250

Time [s]

-10 -15

Fig. 10. Power peak reduction of connected machine tools

In the case of two connected machine tools, the method of exhaustive search [17] can be used as shown in Fig. 10. Finding iteratively the optimized start point of machine 2 by shifting it against machine 1 and checking each scenario. The CPU processing time is still less than 500 ms. For more machines in a connected factory, optimization algorithms like heuristics must be implemented to reduce the CPU processing in future research activities. 4. Quality monitoring through energy data 4.1. Concept of quality monitoring through energy data The value of complex workpieces increases with every process step with the result that a flaw at the end of the production process can be a significant economic loss. By implementing a quality monitoring system in manufacturing, flaws in workpieces can be detected early on. Triggers and reasons for flaws can be identified through the chronological data of monitoring systems [18]. Elaborate fault diagnosis after damage events are prevented and costly analysis and rework

363


Sc Bores

The usability of energy data for condition and quality monitoring purposes was analysed with the following experimental series and settings. As machining process a turning process on a CNC vertical turning center (EMAG VLC100Y) for producing chucked parts was selected. On this machine cylindrical, alloyed heat treated steel (42CrMoS4) workpieces with a diameter of 90 mm and a height of 60 mm were chosen. Workpieces were prepared with bores from 1.2 to 5 mm. As reference part workpieces without preparation were used. The cutting speed was constant, the feed rate was 0.1 and 0.3 mm per revolution. The electrical power demand of the machine drives like axis (x, y, z) and the spindle motor (Sc; nominal power 12.5 kW) was recorded by the EWB with a frequency of 500 Hz. In this experimental setting the recorded data was manual analyzed with Matlab. Fig. 12 (a) and (b) show the power demand of the Axis Sc and z for the reference workpiece. In comparison, the power demand for the workpiece with bores in Fig. 12 (c) and (d) show a significant power drop.

(a)

(c)

150

150 140

130

130

Power [W]

140

120 110 100

90 80

70

70 0

(b)

2

4 6 Time [s]

8

10

Time

Fig. 11. Concept of quality monitoring through spindle power consumption

Another possible scenario are workpieces with material confinements. The material properties of workpiece and confinements are close to each other but still different. If the material of the confinement is harder than the workpiece material an increase in power consumption should be observable, otherwise a decrease. So the basic concept is that drops or increases in power consumption can indicate workpiece flaws. But this also means that comparable data of flawless workpieces must be available either through machining of other workpieces (serial production) or energy simulation based on the NC-Code [24].

(d)

14

14

12

12

10

10

8 6

0

2

4 6 Time [s]

8

10

0

2

4

8

10

8 6 4

2

Tool

100

80

2

0

Spindle Power

110

90

4

Workpiece

120

Power [kW]

Fig. 11 shows the basic concept. The tool is cutting through the rotating workpiece. As the tool reaches the bores the spindles power consumption is dropping as long as the tool is cutting through air and not workpiece material. This is due to the missing drag of the material. The higher the cutting speed and the smaller the bores are, the smaller and shorter the power drop is.

4.2. Experimental settings and results

Power [W]

Workpiece flaw detection To detect workpiece flaws high frequent energy data, in this case provided by the EWB, is necessary. In the following experimental series a scenario for a turning machine was considered: brittle material with break-offs before or during the turning process, simulated by bores.

In the following chapter the experimental series to evaluate the concept for brittle material with break-offs is shown.

Power [kW]

are reduced. By this the safety, reliability and profitability of products are increased [19]. Monitoring systems can be categorized in direct and indirect measuring systems. Direct systems are collecting the relevant data itself, while indirect systems use correlated data [20]. For example to measure the cutting force of a turning machine, it is either possible to use a load cell to directly measure the force or to measure the power consumption (or the current) of the spindle, which is correlated to the cutting force. In turn the spindle’s power consumption is the product of the spindle torque and rotation (plus power losses). Therefore spindle power consumption, torque and rotation are correlated and can be used for different monitoring purposes [21] [20] [22]. Especially in machine tools indirect monitoring systems are utilized. This is due to the fact that sensors are exposed to harsh environmental influences in machine tools: high accelerations, moving parts blocking or disturbing signals, the influence of chippings and coolant possibly leading to sensor damages and therefore to monitoring failures. Apart from these influences, additional costs for components (sensors etc.) have to be considered [23].

0 0

2

4

6 Time [s]

8

10

6 Time [s]

Fig. 12. Power demand of z-axis (a) and the spindle (b) for the reference workpiece and the workpiece with bores (c, d)

An optimization of the recorded raw data is the application of low-pass filter and minimum curve (Fig. 13). This data processing leads to an optimized detection of workpiece flaws. This experimental series has demonstrated the usability of energy (power) monitoring for detection of material break-offs. The implementation of automatic analysis with machine learning algorithms will be one topic of future research.

364


380

360

360

Power [kW]

(b)400

380

Power [kW]

(a)400 340 320 300 280

[8]

340 320

[9]

300 Reference workpiece Workpiece with bores Minimum curve

280

Reference workpiece Workpiece with bores

260

260 1

2 Time [s]

3

1

2 Time [s]

3

[10]

Fig. 13. Power demand spindle raw data (a) and optimized data with application of low-pass filter and minimum curve (b)

5. Conclusion [11]

In this paper, a software framework is presented, which combines a workflow concept with a NC-correlated data acquisition framework based on internal sensor data to optimize the energy efficiency and quality in production. This can be improved, modularizing the machine tool into peripheral devices. Therewith flexible optimization is possible. Power peak reducing algorithms which are to be developed respecting the boundaries can be used similar in a modular machine tool. Condition and quality monitoring systems based on energy data open up new perspectives. The value of energy data increases. Correlated with NC data there is the possibility not only to detect certain workpiece flaws, but also categorize and locate them. With a common software framework it will also be possible to highly automate this system.

[12]

[13]

[14]

[15]

[16]

Acknowledgements The authors are grateful to the German Federal Ministry for Economic Affairs and Energy (BMWi) and the state of Hessia for funding the presented work in the project ETA-Factory.

[17]

[18]

References [19] [1]

[2]

[3]

[4]

[5]

[6] [7]

Freedman, A. 2015. There is less than a 1-in-27 million chance that Earth's record hot streak is natural. http://mashable.com/2015/01/16/ 2014-earth-warmest-year-not-random/. Accessed 8 July 2015. Jan Beier, Sebastian Thiede, and Christoph Herrmann. 2015. Increasing Energy Flexibility of Manufacturing Systems through Flexible Compressed Air Generation. Procedia CIRP 37, 18–23. Menz, B. Energiemanagement: Monitoring, Datenerfassung und Analyse in der ETA Fabrik. Eine Industrie 4.0 Anwendung. In Forum Maschinenkommunikation. Menz, B. and Schraml, P. 2015. Die ETA-Fabrik. Implementierung eines Energiemonitoringsystems. 8. Fachkonferenz Die Energieeffiziente Fabrik in der Automobil-Produktion, München. Abele, E. 2013. Energy-efficient factory of the future – the ηResearch-Factory. 12th Powertrain Manufacturing Conference, Darmstadt. Harbor Research. 2014. What exactly is the Internet of Things? Bundesministerium für Bildung und Forschung. 2014. Die neue Hightech-Strategie.

[20]

[21]

[22]

[23]

[24]

Beck, M., Schraml, P., Menz, B., Helfert, M., Sielaff, T., and Abele, E. 2015. Industry 4.0 Enabler for linked Energy Systems in Producion Sites. In 20° Seminário Internacional de Alta Tecnologia, Inovações Tecnológicas na Manufatura, 06 de Outubro de 2015, Brasilien, ISSN: 2175-9960, K. Schützer, Ed., 64–101. Eoin O'Driscoll, Kevin Kelly, and Garret E. O'Donnell. 2015. Intelligent energy based status identification as a platform for improvement of machine tool efficiency and effectiveness. Journal of Cleaner Production 105, 184–195. C. Herrmann, S.-H. Suh, G. Bogdanski, A. Zein, J.-M. Cha, J. Um, S. Jeong, and A. Guzman. 2011. Context-Aware Analysis Approach to Enhance Industrial Smart Metering. In Glocalized Solutions for Sustainability in Manufacturing, J. Hesselbach and C. Herrmann, Eds. Springer Berlin Heidelberg, Berlin, Heidelberg, 323–328. DOI=10.1007/978-3-642-19692-8_56. Zampou, E., Plitsos, S., Karagiannaki, A., and Mourtos, I. 2014. Towards a framework for energy-aware information systems in manufacturing. Computers in Industry 65, 3, 419–433. Vijayaraghavan, A. and Dornfeld, D. 2010. Automated energy monitoring of machine tools. CIRP Annals - Manufacturing Technology 59, 1, 21–24. OPC Foundation. OPC Unified Architecture. Vorteile und Nutzungsmöglichkeiten für Hersteller und Anwender von Automatisierungs-, IT- oder MES-Produkten. Abele, E., Panten, N., and Menz, B. 2015. Data Collection for Energy Monitoring Purposes and Energy Control of Production Machines. Procedia CIRP 29, 299–304. Menz, B. 2015. Energiemanagement: Monitoring, Datenerfassung und Analyse in der ETA Fabrik. Forum Maschinenkommunikation. Eine Industrie 4.0 Anwendung. http://www.sercos.de/downloads/forummaschinenkommunikation-pr%C3%A4sentationen. Accessed 30 August 2016. Bosch Rexroth AG. IndraMotion MTX efficiency workbench. Zykluszeit und Energieverbrauch reduzieren - Bosch Rexroth Deutschland. http://www.boschrexroth.com/de/de/trends-und-themen/ energieeffizienz/fabrikautomation/loesung-indramotion-mtxefficiency-workbench/loesung-indra-motion-mtx-efficiencyworkbench. Accessed 6 October 2015. Burke, E., Ed. 2014. Search methodologies. Introductory tutorials in optimization and decision support techniques. Springer, New York NY. Kals, H. J. J., König, W., Tönshoff, H. K., Wulfsberg, J., and Luttervelt, C. A. van. 1988. Developments and Trends in Monitoring and Control of Machining Processes. CIRP Annals - Manufacturing Technology 1988, 37/2, 611–622. Grosch, J., Ed. 2014. Schadenskunde im Maschinenbau. Charakteristische Schadensursachen, Analyse und Aussagen von Schadensfällen ; mit 14 Tabellen. Kontakt & Studium 308. expertVerl., Renningen. Wang, L. and Gao, R. X. 2006. Condition Monitoring and Control for Intelligent Manufacturing. Springer Series in Advanced Manufacturing. Springer-Verlag London Limited, London. Tlusty, J. and Andrews, G. C. 1983. A Critical Review of Sensors for Unmanned Machining. CIRP Annals - Manufacturing Technology, 32 (2), 563–572. Gontarz, A., Hampl, D., Weiss, L., and Wegener, K. 2015. Resource Consumption Monitoring in Manufacturing Environments. CIRP Annals - Manufacturing Technology, 26, 264–269. Weck, M., Plapper, V., and Groth, A. 2000. Sensorlose Maschinenzustandsüberwachung. VDI-Z integrierte Produktion 142, 6. Eisele, C. 2014. Simulationsgestützte Optimierung des elektrischen Energiebedarfs spanender Werkzeugmaschinen. Dissertation, Technische Universität Darmstadt.