Building MES-applications with S95

Transcription

1 Presented at the World Batch Forum European Conference Mechelen, Belgium October Fox Valley Drive, Suite 204 Longwood, FL Fax: Building MES-applications with S95 Presented at the European World Batch Forum, October 2004 Ing. Erwin Winkel MBA BAT Manufacturing BV Kerkstaat AA Zevenaar Netherlands Tel: +31 (0) KEY WORDS MES, Manufacturing, S95 ABSTRACT In 2002 at BAT Manufacturing MES applications have been developed around an Oracle RDMS. The MES applications are used to provide operators with near real time information about their production process (the making and packing of cigarettes). The database model is (partially) based on the S95 object model. For the interfaces to other information systems the S95 interface definitions are used. The messages between systems are based on XML and the WBF XMLschema s (as much as possible). The technical architecture is based on the Oracle Application Server and Java (based on the J2EE guidelines). The user interface is completely web-based. The modules realised so far are: 1. Material control. This is done by scanning each material with a barcode scanner before the product is used on the machine; 2. Quality control. Automatic capturing of test-data and displaying at the machine with appliance of SPC-rules; 3. Planning control. Automatic capturing of real-time production-data and verification against planning and shift targets. Copyright 2004 World Batch Forum. All rights reserved. Page 1

2 More modules are planned. For analysis and reporting a manufacturing data warehouse is built, again based on an Oracle-database. Reporting is done with Business Objects. Introduction. In modern manufacturing processes the usage of automated systems is inevitable. Whether for production control, scheduling, or reporting, computers are used. The capturing of data out of the production process and transforming these data into information is one of the objectives of MESsystems. This paper describes the recent development and introduction of several MES-modules in a manufacturing plant. The role of the S95 standard in the project will be highlighted. After a short introduction of the production process, the business goals for the project are briefly discussed. Then the functional and technical architecture will be presented. The results and the benefits of the project will be discussed as well as the benefits of using the S95 standards during the project. The paper ends with the lessons learned. Problem Description Providing information on a shop floor encounters several problems. (1) The information needs to be timely in order to let the operators be able to adjust the process, (2) it needs to be (very) accurate because decisions about the process need to be taken immediate, and (3) it needs to be presented in an operator-friendly way. These requirements place strong demands on the ITsystems in use. When systems are tied to the process itself this is further complicated by the demand for availability. Because of the variety in manufacturing processes it is extremely difficult to implement commercially of the shelf packages. The solution described in this paper is to develop a MES-environment but make use of standard models and concepts. In this way flexibility and extendibility as well as easier maintenance of the system are built in. Project Context The MES-system described in this paper has been built at a manufacturer of cigarettes. The factory does produce approximately 27 billion cigarettes / year. The production process consists basically of two steps: 1) Processing the raw tobacco leafs and blending; 2) Making and Packing the cigarettes. Step 1 is a typically batch oriented process. The blend mixtures consist of 8000 kg batches that are processed in approximately 10 hours processing time. This process is done in the Primary Manufacturing Department (PMD). After the tobacco blends are ready the tobacco is used for making & packing the cigarettes (step 2). The making & packing is done in the Secondary Manufacturing Department (SMD). This is a typical discrete manufacturing process. Based on customer orders and/or existing stock level production runs are scheduled at production lines. Currently BAT Manufacturing has 40 making/packing combinations to produce production runs. The MES-system that is discussed in this paper is situated in the SMD and throughout this text it is referenced as SecIS (Secondary Information System). Copyright 2004 World Batch Forum. All rights reserved. Page 2

3 Project Approach MES-systems typically are situated between the ERP (business) systems and the PCS (production control systems). See Figure 1 for a visual representation of this. One of the challenges in a MES-environment is to build the interfaces required to exchange data between the different layers. The SecIS-project started in 2000 with the drawing up of an inventory of the existing solutions available at other BAT manufacturing plants in Europe. Since these solutions were all tailored to the local situation they could not be easily re-used at the location in Zevenaar. The decision was then made to start with a new implementation in the Zevenaar-plant. First the functional requirements for the MESsystem where defined. This was done with making use of the S95 standard. The objects from Part 1 and 2 of the standard where used as reference in defining the functional requirements 1. Based on the functional requirements a shortlist of the existing commercial available packages was investigated. This was done in At that time the commercial packages did not fit the requirements and therefore a new decision was made to build a new MES-environment (SecIS) at the location in Zevenaar. ERP (Business Applications) MES (Process-Information) PCS (Process Control Systems) The following functional modules where defined: 1) Performance control. This module captures machine data directly at the production machines. Data on downtime, performance, yield etc. would become available for online reporting and feedback to the operator. 2) Quality control. This module captures the quality measurements data. The data then can be used for statistical process control techniques as well as problem and trend analysis. 3) Planning control. This module brings the scheduled production orders online and matches actual production against it. At any moment in time the state of the order can be examined. 4) Material control. All materials are applied with an (EAN128)-barcode containing the article-number. When they are used at the machine they are scanned by the operator and verified against the manufacturing bill. Wrong materials are identified and the operator is warned that something is wrong. This module contributes to the first time right principle. 5) Waste control. During the process waste occurs. The target is to eliminate as much waste as possible. By measuring the waste per production line and reporting on it online visibility for the operator is greatly enhanced. After the functional requirements had been defined, the technical architecture was chosen. The current existing technical architecture as well as strategic corporate guidelines where taken into account when making choices about the technical architecture for the SecIS-environment. The current environment is based on: - Web technology; - Oracle database and Java (J2EE); Figure 1: Automation pyramid in a manufacturing environment 1 At that point in time, Part 1 and 2 were still in draft (SP95). Part 3 of the standard did not exist. Nowadays Part 3 would be used for this step. Copyright 2004 World Batch Forum. All rights reserved. Page 3

4 - A message-queuing communication infrastructure (WebsphereMQ); - (open) standards as much as possible. The following business goals were defined for the project: Empowerment of the shop floor level. In order to empower the operators and let them to be able to make decisions they need information. Therefore SecIS is a state-of-the-art information system primary target at the operators. Development of people. People should develop from machine operator to process operator. For this to happen the operators need to have information about the whole process rather than only the condition of their own machine. Efficiency & Effectiveness improvements. Of course the project did define some quantitative improvements as well. Improvements on the quality of the product as well as the performance of the machines where specified. SecIS provides (near) real-time information on Output, Quality-, and Performance parameters. This yields to the following improvements: Less non-conformity to the manufacturing bill and therefore less faulty products; Enable Statistical Process Control and therefore better accuracy of the product; More precise production to schedule and therefore less waste; No longer need for error prone, administrative work at the machine; Preparation for a faster changing environment that becomes more and more complex. The environment becomes ever more complex. More different versions of the end product lead to smaller runs. Brand changes increase the risk that something can go wrong. Adequate information systems support the operator to produce good-quality products within reasonable time. Copyright 2004 World Batch Forum. All rights reserved. Page 4

5 Where are we now today? To date three modules are operational in the SMD: - Material control; - Quality control; - Planning control. SecIS has contributed to the following production improvements: Empowerment of the shop floor level: First time real proof that materials are sometimes not within spec. Real proof that the scheduling process needs improvement. Development of people: Improve ways of working / procedures. People start to take responsibility for the whole process. Since this represents rather a cultural change this clearly takes some time to settle in the daily ways of working. Preparation for a faster changing environment that becomes more and more complex: The MES-environment has been proven to be extremely flexible. We did not encounter a situation that could not be dealt with (yet). The results on effectiveness and efficiency are as follows 2 : Quality Improvements due to Additional & More Timely Samples Statistical Process Control Techniques Implemented Tangible improvement: MQI 3 from 81% (January 04) to 83% (August 04) and trend is increasing. Productivity Improvements from Less rework Tangible improvement: # of BOM-incidents: From a monthly average of 8 (January 04) to a monthly average of 4 (August 04) and trend is decreasing. Waste Reductions resulting from Better conformance to targets Awareness Tangible improvement: # of rejects: From 2% (January 04) to 1% (August 04) and trend is decreasing. 2 All tangible improvements mentioned are based on the period from January 04 August 04 3 MQI: Manufacturing Quality Index: A formula to express the overall quality of the SMD. Copyright 2004 World Batch Forum. All rights reserved. Page 5

6 Summary and Lessons Learned MES-systems improve the quality of information needed at the shop floor level in several ways: - Information becomes timely available. This make adequate reactions form the operators possible. - Information is reliable. By capturing the data at the source the reliability of the information is greatly enhanced. - The process becomes transparent. By providing information about the whole process than rather a single machine, it becomes clear how the process is running. This prevents sub optimization and enhances responsibility. Also bottlenecks show up more clearly when the process becomes transparent. The overall advantage is that the process becomes more stable and output is more predictable. From the SecIS-project the following lessons could be learned: 1) MES projects need long decision-tracks. This means that circumstances will change in the mean time! During this project a change in the management team occurred and that led to a re-definition of the project. But also take into account: New production circumstances; New machines that need to be connected. New products / manufacturing bills that differ from anything that currently exists. New focus on different problems; e.g. Quality instead of Performance. New requirements due to external factors: Global guidelines, new environmental rules, etc. 2) MES still needs complex technical architectures Although interfaces and software integration are further developed, it is still complex to integrate all components into one system. Not in the last place because production systems tend to be quite old and do not dispose of the latest technological developments. And: You have to connect it all. This means that for a MES-project: Many different skills are needed; Strong project management is needed; A lot of custom build still necessary; You have to connect it ALL!! 3) Standards do help! Standards used in this project: S95, J2EE, EAN128. They were used as: 1) reference when defining the requirements; 2) object models and their attributes; 3) templates during development. (WBF XML-schema s). Since this was the first S95 project from BAT Manufacturing as well as from the system integrator (ICT Solutions) clearly there were some learning effects. Not all choices made in the beginning are the best possible. Therefore, in 2004, some re-structuring of the applications and design will take place to include the new insights. Usage of standards enhances greatly the communication between different parties involved. Also communication between different systems is made easier when definitions are uniquely defined and clear to everybody. Copyright 2004 World Batch Forum. All rights reserved. Page 6

7 Experiences from the system integrator have clearly shown that the usage of the S95-object models and database schema s have contributed to a faster development process. The development of an e-mes Framework based on S95 will greatly enhance future developments in the MES-environment. 4) Beware of exceptions (everywhere!) Especially in a production environment you will encounter exceptions that you will need to take into account: Special (old) machines; Special manufacturing bills; Special procedures; Special materials Having a MES-environment that is flexible enough to cope with these exceptions is an enormous advantage. Until now we did not encounter a situation that we were not able to tackle. 5) Existing culture is as important as technology!! In a production environment procedures and work-instructions can be quite old. Old habits are difficult to weed out. People need time to learn with a new system and they need to learn to trust the system. That means that a lot of attention is needed in the introduction phase. Therefore, a lot of effort was put into training of the operators (approx. 150 persons), quality specialists (25 persons), and operational management (12 persons). Even then, it proofs that trust is difficult to obtain and the system has to proof itself over a long period before people believe what the screen shows them. Copyright 2004 World Batch Forum. All rights reserved. Page 7