On-line Quality analysis integration in Batch process operations Jean Vieille Consultant Strasbourg, France Philippe Fabre Project Manager Cray Valley Drocourt, France KEYWORDS Batch process Control, Finite Capacity Scheduling, Quality analysis, Material adjustment, Sampling, Summarized process information, LIMS, MES, ERP, FCS, S95.01, S88.01 ABSTRACT This paper exposes a solution developed to assist the Production Quality Assurance business process for a polymerization plant. This solution aims to provide: A dynamic Quality Analysis management synchronized with ERP data and production activities A human-independent process performance A complete and accurate information about the process behavior. Quality analysis may be performed either by the lab team or by the operators themselves. The analyses to be performed depend on the product and on the current process operation. They are often iterative, requiring material additions or set point adjustments. The quality specification requirements are best attached to the ERP finished product and customer data. The relationships between the lab and operation team involve real-time analysis orders management and performance reporting. The exposed solution gives a way to deliver the quality specification requirements linked to the production schedule from the ERP, and to get the quality analysis reports attached to the production performance data to the ERP. The quality specification includes the formula for computing the material adjustments. The samples are considered as sub-operations in the product processing, and considered as work orders in the lab activity. The system provides a tight real-time coupling between production rules, material handling and quality control at the operating level, and a secured asynchronous quality information link between ERP and production system. The new system had not been fully implemented at the time this paper was written. The Quality analysis integration was designed as an add-on to be plugged into the material and operation control functions of the new plant supervision system. True benefits of this project will be reported during the presentation Introduction The CRAY VALLEY Drocourt' facility produces polymer resins by syntheses and blending. The enterprise information system retrofitting based on a new ERP system dictated to reconsider the production driving under several points of view: ERP and Control system production rules synchronization Packing / Production synchronization Resources usage optimization Material inventory management Detailed production information Quality Analysis management The subject of this presentation is mainly concerned with the quality control in production and its integration through the Business / Manufacturing boundary. Although it was specified before the availability of this standard, the data model used could illustrate a true-life example of ISA S95.01 standard application. Integration is not a goal by itself. As far as the quality control is concerned, the true goals were: Providing a flexible and consistent Quality management which allows last time changes of Product and Customer Quality requirements Recovering the human process knowledge and tracking the process behavior Getting a predictable plant performance, reaching the requested specifications in the same way whoever the operating team is. The following considerations led to the exposed solution: The information system must be consistent between Product / Customer quality requirements, analysis process and quality results, allowing an automated Production Quality Assurance business process The Quality specification requirements are best attached to the ERP product and customer data. Quality analysis may be performed by the lab team or by the operators themselves. The relationship between the lab and the operation teams involves real-time analysis orders management and performance report. The analyses to be performed depend on the product and Customer specific requirements and on the current process operation. They are often iterative, requiring material additions or set-point adjustments to get the target or to follow the typical evolution of the specifications The actual LIMS was out-of-date and had to have been retrofitted This solution is characterized by the following: Quality specification requirements are linked to the production schedule delivered by ERP Quality performance reports are attached to the production performance data transferred to ERP Computing formulas for material adjustments are part of the Quality specifications Samples are considered as sub-operations in the product processing, and are managed as work orders in the Lab activities. Information about past equivalent productions is available (Analysis results and material adjustments). This allows comparison and likelihood control against system suggestions. The system provides a tight coupling between production rules, material handling and Quality control, and a loose, secured coupling between ERP and production system Analysis and Quality Control - Scope of the presentation The control and analysis activities may be classified in several domains: Raw material stock and receipt controls On-line, End-of-production controls Packing and shipping controls Finished products stock periodic and obsolescence controls In our case, the domains 2 and 3 are subject to strong time constraints and act directly on the main processes they assist. Domains 1 and 4 are generally less critical. Raw material receipt control was excluded from the project scope, but could have been managed by the same way without any particular difficulties. The production is delivered either in bulk or in packaged items. Shipping of packed product is buffered and does not need further controls, while bulk delivery may be synchronized with production and needs specific controls. Finished products stock Control has been integrated, but will not be developed here. On-line and End-of-Production Controls Quality Analysis of a product during its processing, combined with the other process data, helps to: Master its characteristics at different stages Match the final specifications Improve the process behavior knowledge We notice that: Input materials (raw and semi-finished) specs deviations, operating conditions and production rules disrespect lead to deviations of the product expected characteristics. These deviations must be corrected in order to match the expected product specs for the next process stage, or the finished product specs. The corrections are mainly material compensation in our case, The production efficiency requires operation duration optimization as well as fast and suitable actions to compensate the deviations The lab activity is supposed totally responsive to the production needs, although this department must assume analysis orders coming from other units and should arbitrate its priorities. The analysis work may be completed by one or several labs, and sometimes by operators themselves. The work may be shared between different actors: sampling, analysis, material correction, and quality status assignment. Process overview There are 2 main processes by which the plant produces polymer: Synthesis from raw materials Blending of semi-finished products and raw material The method adopted and presented here was fully suitable for theses 2 type of processes, as well as for the stock control and adjustment considered as "production order" and included in the ERP production schedule. As shown bellow we can see a production order linked to one shipping order (there may be several shipping / packing orders linked to a single production order) Synthesis process: Production segment / QA activities for production and synchronized shipping Production Operations Shipping operations Production segment Reactor loading Reaction + Spec adjustement Diluter transfer + loading Spec Adjustment Control before conditioning Conditioning QA activities - On-line control by Operators - On-line control by Lab Control By Lab Control by Lab Equipment requirements Reactor Reactor Reactor diluter Diluter Diluter Conditioner Conditioner Personal requirements Operator Operator Operator Operator Lab assistant Operator Lab assistant Operator Lab assistant Material requirements ERP Formula Analysis results based computed adjustment ERP Formula Analysis results based computed adjustment - - Information System architecture The production information system is built around 3 sub-systems: ERP, DCS, MES. The ERP provides mastering of "processes" which define production segments including basic formula, equipment class requirement, personal requirement and QA specs. These processes can be seen as something between the Master Recipe and the General/Site Recipe, taking account of the resources requirements, but summarizing the procedural elements. The DCS runs an S88 batch engine. The recipes are built to represent a detailed view of the ERP' processes. The recipe execution is synchronized by the DCS, which sends flags towards the ERP/MES high level "site" recipe. In order to comply with the company's actual control system safety requirements, no data are passed automatically from the ERP. Batches are initiated and launched manually (10 batches per day). Some units are manually driven. The "MES" is a communication-enabling framework with complementary applications. It manages the data exchange and applications monitoring. It was specifically developed, by strategic choice rather than by technical reasons. Applications are generally custom-made. As an exception, the finite capacity scheduler is an off-the-shelf software. This layer was designed to be fully "transparent", offering only security access and environmental configuration. Operational data mastering is always dedicated to its owner (originating application), which is ERP in most of the cases. This is a duty to guarantee the system integrity and to reduce its maintenance. The DCS "pushes" the process data into the production database in real time or at the end of the batch depending of the system requirements. Production Driving Initial production schedule from ERP The ERP's schedule contents all released production orders. A production order is made of: Production scheduling information. It includes the production plan (what to produce, when, how) and links between production requests, for instance packing and production orders Production definition information. It includes production rules, detailed material and QA requirement, class-based equipment and personal requirement. It is split into Product Segments according the production rules (process stages and operations). The production rules included in the schedule are interpreted by the FCS using predefined production steps. The Production Capability information is shared between the ERP and the FCS (Finite capacity scheduler): Material capability is directly managed by ERP. It maintain a list of usable lot ID, location, packing and quantities for each requirement Equipment and Personal capability is managed by the FCS 2 production schedules are managed separately and contained into 2 separated databases: The Current Production Schedule The Simulation Schedule, which is used to check the overall facility load against the real capacity for short, medium and long term ERP scheduling and forecast, involving the FCS constraint analysis functionality. This model does not fit exactly with the current S95.01 model (draft 11 June 1999): Material requirement is supported by a single object rather than being separated in "Produced", "Consumed" and "Consumable" materials objects. A separate object supports QA requirement, rather than being encapsulated in material model. This design was preferred because several production requests for the same product may involve different QA requirements according the specific customer needs. The S95.01 model could have been applied as well, but in a less obvious way. Actual schedule: Requirement completed by Controls and FCS The subject of this paper is exposed here. The production segments are split into sub-segments generated for each sample. This allows a detailed production information to be managed at the sample level through a very simple schema. The sampling is managed according the QA requirements. The material adjustments recommended after the sample analysis according the QA requirements are introduced in the material requirement information, which is dynamically expanded to the "sub-operation" sample level. In addition to QA material alteration, this model supports formula alteration to take account of unexpected material loading (due to operator feeling, or lack of requested material, or request from QA...). The finite capacity scheduler translates the class-based Equipment requirements into specific allocations ("Reactor" in the initial schedule becomes "R151" after FCS operation). It defines the Personal requirements. Personal and Equipment resources will not be longer detailed here, being outside of the presentation scope. The material lot tracking and resources usage are not detailed here. Production performance and reporting to ERP The production information is recorded into the Current Production Information database during the production execution according a model symmetric to the previous one. At the end of the production of a particular production request, the production information is transferred into a separate Historical Production Information database. This database may be used: On-line as reference to actual process behavior, by chemists or operators Off-line for process behavior analysis, material transfer inconsistencies enquiry, equipment usage computation, Pareto failure analysis... At the same time, a production performance report is sent to the ERP. It consists of a consolidated and validated information set built from the detailed production information. This information is summarized according the ERP's needs: Resources activities: Resource allocation and duration of each production segment (only equipment usage is reported) Material transaction: Summarized material usage and production by lot, location and category QA results: Consolidated QA results for the finished product. It includes the test results as well as the quality status for the produced material. We get a very simple information response to the ERP's production request, while the detailed information remains available in the production information database. Again, this model does not fit exactly the S95.01 model: QA results are reported in a separate object (corresponding to the schedule) QA and material are reported against the schedule, not against the production segment. Quality Control Driving Quality Control operational scheduling The lab activities are managed in the same way than the production facility. In fact, a Lab is equivalent to a Process Cell and samples are considered as analysis orders in an analysis schedule. By this way, the main laboratory and the operators performing particular on-line auto-control analysis get their own independent schedule. The production schedule, production segment and material information are hidden (but available) in the analysis schedule, providing an integrated, but independent activity management. The analysis schedule should be matched by the analysis performance (current data) and may refer to the past performances for the same product and (optionally) the same equipment. This allows an easy comparison between the actual results and material adjustments and the past production sessions. Using the same finite capacity scheduler than the equipment allocation may refine the Labs activity. Conclusions The presented system offers a suitable but simple model to integrate the Lab information into the production information. The solution developed upon coached user specifications corresponds nearly to the needs expressed by the production department while complying with the ERP production information gathering. A complete and accurate information is available regarding the material formula adjustment. Each material addition may be compared to the corresponding physical and chemical properties of the product for the corresponding process operation and QA sampling The required summarized information is return to ERP while keeping the detailed information at the production level The lab work is managed efficiently, a finite capacity scheduler may be implemented easily The S95.01 model is not fully implemented and respected. However, specified in the ignorance of this standard, this solution is not so far of it and could have been developed in accordance with its conceptual models. The main benefit against the past LIMS system is the efficient dialog between the Production Plant and the Lab and the ERP-centric QA management, which simplifies dramatically the product data management. Conversely, we could mention some deficiencies: Mainly custom-made, this MES architecture could have been built using off-the-shelf components. This choice has led the project to uncontrolled delay issues, technical uncertainties and budget / cost deadlock either for the final user and the contractors. The system is not actually a full integration example, leaving the DCS system still isolated regarding the top-down information flow. The responsibility sharing issue between Business and Control system has not been solved. This is a main challenge for SP95! The S88.01 standard was never considered (except at the DCS level). The physical model is managed independently in 3 different systems (ERP/FCS/DCS). Thus, the main issue of this project is its uncontrolled schedule due to endless system specification and custom development. If it had to be redone, its implementation would have been considered in a more efficient way: Use commercial packages from reliable vendors as much as possible. The company, which developed the system components, disappeared before the end of the project. The ERP provider had to take the responsibility to finish and support the system. The DCS system was left out of the project. Because of unclear DCS' vendor strategy or misunderstanding, its batch management capabilities were not taken into account. The operators have to use 2 different packages and the synchronization with the actual process state is still manually managed. A better choice would be to consider an extensive usage of the DCS batch management capabilities, even for manually driven process cells. A much stronger project team and a more realistic budget would probably lead to better options and project schedule control. We hope than the basic concepts exposed in this presentation would give valuable inputs to many Business/Production systems integration projects. Revised DATEENREG \@ "MMM-aa" \* FUSIONFORMAT Jul-99 Packing, Shipping Control Shipping Finished product Stock Control Packing Production Raw mat. Stock, Receipt Control On-line, End Prod Control Receipt Stock Scope of this presentation LAB 1 Production unit A Quality Assurance LAB 2 Production unit B Sampling, Specs QA status request, Specs QA status assignment Auto-control Results, Corrections Perturbations Production Schedule QA specs Materials Requirement Equipment Requirement Production Performance QA Results & Status Material Actual Equipment Actual Reactor Diluter Raw materials Tuck loader Production Shipping ERP : Company Management Information, Production Scheduling MES - Intra-Area Coordination DCS - Control system with batch execution system Finite capacity Scheduling QA control Production control (operations, materials) Production and analysis orders Manager Data exchange monitoring Production information database Production Schedule 0..n Material adjustment request Generate values for Correspond to 0..n 1..1 Production Request 0..n 0..n 0..n 0..n 0..n 0..n 0..n 0..n Production Request Segment Requirement Class-based Equipment requirement Material requirement QA requirement 0..n May be linked to 0..n 0..n 0..n Provide computational information for 0..n 0..n QA Actual Material Actual Material Requirement (actual) Analysis Request Formula adjustment rules Equipment Actual QA requirement Material requirement (initial) Equipment and personal requirement Segment Requirement Production Request Production Schedule Segment Response Production Response Production Performance Correspond to May refer to 0..n Correspond to QA test specification 1..1 Unexpected material usage 1..1 1..1 0..n Production Response 0..n Material requirement (initial) 0..n May be reported on by QA test specification Production Response Material adjustment request 1..1 0..n Analysis Request (samples) Analysis Schedule May be linked to May be reported on by 1..1 Formula adjustment rules Refers to 1..n May contains 1..n 1..1 0..n 0..1 QA test Results Material adjustment request Analysis Response Analysis performance 0..n Raw materials QA test Results QA test Results Analysis Response Analysis history 0..n May correspond to May correspond to Material produced, Equipment used 1..1 0..n Material adjustment actual 0..n Material adjustment actual May correspond to May correspond to