Collaborative and Distributed Chemical Engineering. From Understanding to Substantial Design Process Support

The following section discusses the

long-term approach

of IMPROVE. We start with a careful analysis of current design processes in chemical engineering and their insufficient support by tools. Then, we sketch how new ideas from the engineering and informatics side can considerably improve the state-of-the-art. The resulting tools according to these new ideas are sketched in the next section.

It should be remarked that both, the long-term goals and the principal approach of IMPROVE [343], were not changed within the whole project period, i.e. – if also counting preliminary phases of the project – within a duration of 11 years. This

stability

of

goals

and

approach

shows that we have addressed a fundamental and hard problem which cannot be solved in a short period. It, furthermore, shows that the undertaken approach went in the right direction. Therefore, it was not necessary to revise the approach within the whole project period, neither in total nor in essential parts.

The IMPROVE demonstrator is an integrated research prototype of a novel design environment for chemical engineering. Throughout the IMPROVE project, it has been considered essential to evaluate concepts and models by building innovative tools. Moreover, an integrated prototype was seen as a driving force to glue research activities together. Two versions of the demonstrator were implemented. The first version was demonstrated after the first phase of the project (from 1997 to 2000). The second demonstrator built upon the first one and was prepared at the end of the second phase (lasting from 2000 to 2003).

This section describes the second demonstrator, which shows interdisciplinary cooperation (between chemical and plastics engineering), interorganizational cooperation (between a chemical engineering company and an extruder manufacturer), and synergistic tool integration.

As already introduced in Sect. 1.1, IMPROVE is a

joint project

between

different disciplines

, namely Chemical Engineering and Informatics. Even more, Plastics Engineering and Labor Research are also involved. Informatics itself is represented by three groups of different research fields.

Whereas Sect. 1.1 dealt with the problem to be solved and Sect. 1.2 discussed a resulting integrated prototypical environment, this section discusses the project and the

research and development process

to solve these problems.

The

section

is structured as

follows

: First, the IMPROVE project structure is discussed, next its iterative and practical approach is introduced. Two further characterizations are given, namely the funding of IMPROVE and the predecessor projects. Finally, we give a survey of this book’s structure.

This section serves as an introduction to application domain modeling. Firstly, we will motivate the objectives of modeling. Next, we will suggest definitions for different types of application domain models. Finally, a brief survey of the modeling languages applied throughout this chapter will be given.

This contribution summarizes the results of more than a decade of product data modeling at the authors’ institute. In this effort, a series of consecutive product data models has been developed for the domain of chemical engineering, starting with the chemical engineering data model VeDa, followed then by the conceptual information model CLiP and finally by the domain ontology OntoCAPE. The evolution of these different models is described, and their similarities and differences are discussed with regard to scope, structural design, conceptualization, representation language, and intended usage.

In this contribution, a comprehensive document model is presented, which describes the types and dependencies of documents as well as their internal structures and dynamic behavior. Additionally, the contents of documents can be indicated by integrating the document model with a product data model. Due to these properties, the document model provides an adequate basis for a novel class of software tools that support efficient document handling in development processes.

Empirical studies are a prerequisite for creating meaningful models of work processes, which can be used to analyze, improve, and automate design processes. In this contribution, a modeling procedure is presented, which comprises the creation of semi-formal models of design processes, their analysis and improvement, and finally the formalization of the models as a prerequisite for the implementation of supportive software tools. Several modeling languages have been created for representing design processes, including the C3 language for participative modeling of design processes on a semi-formal level and a Process Ontology for the formal representation of design processes.

In this contribution, the last of the four application domain models is presented. The model, implemented as an ontology, covers the rationale underlying the decisions in design projects. The ontology supports the representation of concrete decisions for documentation purposes as well as generalized decision templates, which can serve as guidelines for designers and help to reduce the effort for documenting concrete decisions.

A comprehensive summary of the application domain models presented in this chapter is given, and their integration into a common framework is discussed. Other existing application domain models of comparable scope are reviewed and compared to the models presented herein.

The process industries are characterized by continuous or batch processes of material transformation with the aim of converting raw materials or chemicals into more useful and valuable forms. The design of such processes is a complex process itself that determines the competitiveness of these industries, as well as their environmental impact. Especially the early phases of such design processes, the so-called conceptual design and basic engineering, reveal an inherent creative character that is less visible in other engineering domains, such as in mechanical engineering. This special character constitutes a key problem largely impacting final product quality and cost.

As a remedy to this problem, in cooperation with researchers and industrial partners from chemical and plastics engineering, we have developed an approach to capture and reuse experiences captured during the design process. Then, fine-grained method guidance based on these experiences can be offered to the developer through his process-integrated tools. In this section, we describe the application of our approach on the case study of the IMPROVE project. We first report on experiments made with a prototypical implementation of an integrated design support environment in the early project phases, and successively describe how it has been reengineered and extended based on additional requirements and lessons learned.

Design processes in chemical engineering are inherently complex. Various aspects of the plant to be designed are modeled in different logical documents using heterogeneous tools. There are a lot of fine-grained dependencies between the contents of these documents. Thus, if one document is changed, these changes have to be propagated to all dependent documents in order to restore mutual consistency.

In current development processes, these dependencies and the resulting consistency relationships have to be handled manually by the engineers without appropriate tool support in most cases. Consequently, there is a need for incremental integrator tools which assist developers in consistency maintenance. We realized a framework for building such tools. The tools are based on models of the related documents and their mutual relationships. Realization of integrators and their integration with existing tools is carried out using graph techniques.

The development of design processes in chemical engineering and plastics processing requires close cooperation between designers of different companies or working groups. A large number of

communication relationships

is established, e.g. for the clarification of problems within single tasks, or within project meetings covering the discussion about interim results. With the ongoing development of a process, different types of communication relationships will occur, as the required communication form as well as the extent of communication are depending on the work task. For an efficient support of the communication between designers, support tools are needed to enable cooperative computer-aided work tailored to an actual task and to obtain a speed-up of the development of a process.

This section discusses several activities to improve the communication between engineers by

multimedia and Virtual Reality tools and protocol mechanisms

for supporting

new forms of cooperative work

in the design of a process. To ease the usage of those tools and protocols within the work processes, they are integrated into the normal working environments of the designers. For communication processes involving geographically distributed designers, the

communication platform KomPaKt

was developed, which integrates the new communication and cooperation tools for different purposes in a single, configuration-free, and intuitive user interface. As a specially interesting case of cooperation between designers, the technology of

immersive Virtual Reality for simulation sciences

was examined in more detail using the example of compound extruders, as Virtual Reality technology plays a key role in interdisciplinary communication processes.

Design processes in chemical engineering are hard to support. In particular, this applies to conceptual design and basic engineering, in which the fundamental decisions concerning the plant design are performed. The design process is highly creative, many design alternatives are explored, and both unexpected and planned feedback occurs frequently. As a consequence, it is inherently difficult to manage design processes, i.e. to coordinate the effort of experts working on tasks such as creation of flowsheets, steady-state and dynamic simulations, etc. On the other hand, proper management is crucial because of the large economic impact of the performed design decisions.

We present a management system which takes the difficulties mentioned above into account by supporting the coordination of dynamic design processes. The management system equally covers products, activities, and resources, and their mutual relationships. In addition to local processes, interorganizational design processes are addressed by delegation of subprocesses to subcontractors. The management system may be adapted to an application domain by a process model which defines types of tasks, documents, etc. Furthermore, process evolution is supported with respect to both process model definitions and process model instances; changes may be propagated from definitions to instances and vice versa (round-trip process evolution).

The research of the IMPROVE subproject C1 “Goal-Oriented Information Flow Management in Development Processes” aims at the development and evaluation of database-driven methods and tools to support and optimize the distributed storage and routing of information flows in cooperative design processes. The overall concept of a

Process Data Warehouse (PDW)

has been followed which collects, and selectively transforms and enriches required information from the engineering process. The PDW has been conceptually based on interrelated partial domain and integration models which are represented and applied inside a metadata repository. This allows to query and apply experience information based on semantic relationships and dependencies. Special attention has been paid to aspects of cross-organizational cooperation.

Tools used in development processes in chemical engineering and plastics processing are often highly

heterogeneous

with respect to the necessary software and hardware. In the previous chapter, several functionalities for improving a development process in different aspects was presented. But one problem still remains: when coupling such heterogeneous tools, many technical details are to be considered.

This section describes a

service management platform

that aims at

hiding such technical details

from new tools as well as from developers, and at ensuring a

performant execution

of all tools and services. The integrative support presented in this section is located at a lower level. Here, the focus is on the provision of a transparent, efficient, and fault-tolerant access to services within the prototype developed in IMPROVE. A framework was developed that allows efficient communication support, and the management of both, integrated tools and external services. The framework also allows an a-posteriori integration of existing development tools into the management platform. Framework and management are applied at platform level in this section.

In this section, the modeling procedure for design processes introduced in Subsect. 2.4.2 is discussed from a more application-oriented point of view. The Workflow Modeling System WOMS, which has been developed for the easy modeling of complex industrial design processes, is described. Many case studies have been performed during the elaboration and validation of the modeling methodology and the tool, several of them in different industrial settings. In this contribution, some case studies are described in more detail. Two of them address different types of design processes. A third case study, demonstrating the generalizability of our results, deals with the work processes during the operation of a chemical plant.

The design and optimization of creative and communicative work processes requires a detailed analysis of necessary activities, organizational structure, and information flow as well as the identification of weak spots. These requirements are met by the C3 modeling technique, which was specifically developed for design processes in chemical engineering (cf. Subsect. 2.4.4). C3 is also the foundation of the simulation-based quantitative organizational study described in this section. Therefore, a transformation technique from semi-formal models of work organizational dependencies into formal workflow models has been developed and implemented. The verified results of test-runs show the various fields of application of this technique, including its benefits for the reduction of cycle times, for the optimization of the operating grade of the employees, and for the capacity utilization of tools and resources.

The development of chemical processes requires the consideration of different areas such as reaction, separation, or product conditioning from several perspectives such as the economic efficiency of the steady-state process or the performance of the control system during operation. Mathematical modeling and computer simulation have become vital tools in order to perform such studies. However, various applications are supported by a number of tools with different strengths and weaknesses. Unfortunately, the formulation of the mathematical model and the data structures underlying the implementations of these tools are incompatible. Their integration at runtime for chemical process development poses technical problems to the engineer who carries out the work. This section describes an environment that accounts for the differences of mathematical modeling and simulation tools, and facilitates modeling and simulation across the boundaries of incompatible tools. The use of this environment is illustrated in the context of the IMPROVE reference scenario, addressing the production of Polyamide-6 from

ε

-caprolactam (cf. Subsect. 1.2.2); the simulation tools used in this case study are currently used in industrial design processes.

This section describes the different dimensions of integration inside the plastics processing domain as well as cross-organizational integration and collaboration issues. The presented results range from work process modeling up to technical process analysis for the design of compounding extruders in the chemical engineering context. Standard practices for the design of polymer compounding extruders were analyzed and afterwards formalized in cooperation with subproject I1 using methods and tools developed and used in the CRC 476. Fragments of these workflows were redesigned using innovative informatics functionality provided by the CRC’s B-projects which provided the novel tool functionality. Exemplarily, the extruder simulation tool MOREX was integrated with the process-integrated modeling environment PRIME and coupled with BEMflow. The distributed analysis of 3D simulation results using KOMPAKT and TRAMP was developed, and a scenario showing the integration of the project management system AHEAD with the plastics engineering design tools was designed participatively. Another focus was set on an integrated visualization environment using Virtual Reality technology for different data from a number of simulation tools.

The novel informatics concepts presented in Sects. 3.1 to 3.4 can be integrated again to fully exploit their synergistic potential. This section shows three interesting examples of such synergistic integration. The examples bridge different roles or companies in the design process. They also bridge between the efforts of different research groups within IMPROVE.

The number of employees working exclusively with computers increased almost 20 percent within the last four years. In technical offices and in the field of research and development the largest amount, 94 percent, can be found. Hence the focal point of the research project was composed of the development and prototypical implementation of software tools for the support of the work of process development engineers. This field of application provides a large range of innovative computer support because of its high amount of creative work processes hardly to support by strictly structured software tools. Additionally the exploration of many design alternatives and weakly structured constraints between different activities with unexpected or planned iterations can scarcely supported in a conventional way. For this research objective several specialized applications were developed and the ergonomics of their user interfaces were evaluated. These application specific evaluation methods form the groundwork for the redesign based on the design recommendations.

The following section describes the foundations of software ergonomics with corresponding international norms. It will also be shown, that software ergonomics knowledge alone, however, is not sufficient in order to be able to assess the quality of the software support. So the well known ergonomic requirements were supplemented by methods for work analysis. In general, the support of creative and communicative work processes requires a detailed analysis of necessary activities, organizational structure, and information flow including the identification of organizational bottlenecks. The basis for the conducted evaluation of work thus constitute three different models of work analysis which are introduced and discussed.

A suitable procedure was presented, not only for the development of evaluation criteria, but also for application scenarios for the following evaluation of subjects. The scenarios will be briefly described, and the results of the various evaluations of subjects will be introduced. These findings show the quality of the developed prototypes in terms of their area of application and give recommendations for possible improvements.

The a-posteriori integration of heterogeneous engineering tools, where tools are supplied by different vendors, constitutes a challenging task. In particular, this applies to an integration approach where existing engineering tools are extended by new functionality which, again, can be integrated synergistically. Responding to these challenges, an approach to tool integration is described which puts strong emphasis on software architecture and model-driven development.

Starting from an abstract description for an integration software architecture, this architecture is gradually refined down to the implementation level. To integrate heterogeneous engineering tools, wrappers are constructed, abstracting from technical details and providing homogenized data access. This approach to tool integration is supported by a collection of tools for software architecture design and model-driven wrapper development, all based on formal graph models and transformation rules. This collection of tools considerably leverages the problem of composing a tightly integrated development environment from a set of heterogeneous engineering tools. So, we give specific architecture tools for the problem of tool integration following an a-posteriori approach.

This section gives an introduction on the relation of the process/product model to tool development. We concentrate on tools offering novel functionality (experience-based, consistency, reactive, see Sect. 1.1 and Chap. 3). We motivate again the challenging task of developing such a comprehensive model. We give a summary how application domain models are structured (for details see Sect. 2.6), and we present how the corresponding information fits the tool construction process. A categorization of open problems related to our preliminary results for a coherent, comprehensive, and uniform model is given at the end of this section. The problems, themselves, are presented in Sect. 6.5. The section also serves as an overview for this chapter.

The first vertical column of the layered process/product model (PPM) addresses the direct, experience-based support at the technical workplaces of designers. More specifically, we demonstrate the transition from application domain models to executable tool models, focussing on the process perspective of the PPM. This vertical column is jointly realized by the A1 subproject, providing the fine-grained application domain models, and the B1 subproject, dealing with their conversion to executable tool models to be used by process-integrated tools. In this contribution, we provide an outline of the cooperation results.

The models developed within subprojects A2 and B2 together form one of the vertical columns of the process/product model. The application domain models of A2 are refined to tool models of B2 such that integrator tools can be realized. The process of building integrators is rather well understood in general, as is the process of refining the application domain models of A2 to tool models of B2. Nevertheless, important parts are missing for a concise and layered process/product model.

One of the vertical columns in the overall process/product model deals with the cooperation of subprojects I1 and B4. Both study the support for reactive process management in dynamic development processes. In this section we highlight the transition from application models developed in subproject I1 to tool models for reactive management of subproject B4. We summarize our findings w.r.t. the development of management tool models as well as their connections to application models. The section focuses on a process-oriented viewpoint. Products and resources of development processes can be discussed analogously. We identify the missing parts which need to be further investigated in order to get a comprehensive and integrated process/product model, here for reactive management.

In this section we give a summary and evaluation of what we have achieved w.r.t. the PPM. Especially, we discuss the open problems still to be solved. The message is that there are a lot of nice results, but there has also a lot to be done in order to get a uniform and comprehensive PPM.

This short section is to demonstrate that one of the characteristics of IMPROVE was a permanent exchange of ideas with industrial partners: Ideas were taken up from industry, symposia and workshops were held, spin-offs were founded, etc. So, the transfer center described in this chapter is essentially the result of our long-lasting and continuing cooperation with industry.

During the design phase of a chemical plant, information is created by various software tools and stored in different documents and databases. These distributed design data are a potential source of valuable knowledge, which could be exploited by novel software applications. However, before further processing, the scattered information has to be merged and consolidated. For this task, semantic technologies are a promising alternative to conventional database technology. This contribution gives an outline of the transfer project T1, which aims at the development of an ontology-based software prototype for the integration and reconciliation of design data. Both ontology and software development will be performed in close cooperation with partners from the chemical and software industries to ensure their compliance with the requirements of industrial practice. The advantages of semantic technologies will be demonstrated by comparing the prototype against a conventional integration solution.

The transfer project aims at the integrative modeling, analysis, and improvement of a variety of work processes in the life cycle of a chemical product across disciplinary and institutional boundaries. A methodology is elaborated for the creation of conceptual, coarse grained models of work processes originating from empirical field studies in industry and their subsequent enrichment and formalization for computer-based interpretation and processing.

The results of IMPROVE, which are extensively described in this book, are generally interesting for industrial users. In this transfer project, the insights gained in subproject I4 are expanded for the simulation-supported planning of process design and development projects with partners from the chemical and software industries. The planned activities are presented in this section. The goal of the transfer project is the interactive, in parts even automated transformation of work process models – which are created using the participatory C3 modeling method – into workflow models suitable for simulation studies. The required formal representation of workflows is based on Timed Stochastic Colored Petri Nets. That way, the systematic variation of organizational and technological influencing factors and the analysis of organizational effects becomes possible. By providing this simulation method already in the planning phases of design and development projects, precise prognoses for better project management can be achieved.

Extrusion of rubber profiles, e.g., for the automotive industry, is a highly complex continuous production process which is nevertheless influenced strongly by variability in input materials and other external conditions. As analytical models exist only for small parts of such processes, experience continues to play an important role here, very similar to the situation in the early phases of process engineering studied in CRC IMPROVE. This section therefore describes a transfer research project called MErKoFer conducted jointly with an industrial application partner and a software house founded by former CRC members.

In MErKoFer, results from the CRC projects on direct process support (B1, see Sect. 3.1), process data warehousing (C1, Sect. 4.1), and plastics engineering (A3, see Sect. 5.4) were applied and extended. Specifically, knowledge about extrusion processes is captured by ontology-based traceability mechanisms for both direct process support of extrusion operators, and for process analysis and improvement based on an integration of data mining techniques. The accumulated knowledge assists in ensuring defined quality standards and in handling production faults efficiently and effectively. The approach was experimentally implemented and evaluated in the industrial partner’s site, and some generalizable parts of the environment were taken up by the software house partner in their aiXPerience software environment for process automation and process information systems.

The results of the IMPROVE subproject B2 (cf. Sect. 3.2) are to be transferred to industry. The corresponding transfer subproject T5 is described in this section. The main goal is to provide a universal integrator platform for the engineering solution Comos PT of our industrial partner innotec. The core results to be transferred are the integration rule definition formalism and the integrator framework including the execution algorithm.

Besides evaluation in practice, the transfer subproject will also deal with major extensions. For instance, the integration rule definition formalism will be extended and repair actions to restore consistency of damaged links will be incorporated into the framework. The transfer is intended to be bidirectional, i.e. our partner’s knowledge will influence our research as well.

In the past funding periods of the CRC 476 the process management system AHEAD has been developed as a research prototype (cf. 3.4). The project T6 aims at transferring the corresponding research results into two different industrial environments. Together with innotec GmbH, we realize a process management system on top of the chemical engineering tool Comos PT. This system will allow for the holistic management of the overall administration configuration (activities, products and resources). Furthermore, we extend our application experience by also considering dynamic business processes. In cooperation with AMB-Informatik, we build a workflow management system that allows dynamic changes of workflows at runtime. This new workflow management system is also built on top of an existing one, which strictly separates build-time from runtime.

Service-oriented architectures define an architectural style for the construction of a heterogeneous application landscape. By abstracting services, business processes are decoupled from the underlying applications.

This section describes how the results of the IMPROVE subproject I3, related to model-driven development process for wrapper construction, are transferred and extended within the area of business applications. We present an approach which yields a prototype to formally specify service descriptions and service compositions. This prototype makes it possible to evaluate and explore service-oriented architecture concepts.

This section briefly summarizes and evaluates the results of IMPROVE from an application-oriented perspective. The application domain model, its use for the improvement of design processes, and its implications for the development of design support tools constitute the main focus. We conclude that the model framework has reached a high standard and constitutes a good basis for further refinement and extension. However, the model needs even more validation in industrial case studies. The discussion of this section is given from an academic viewpoint. The industrial relevance of design process results is given in Sect. 8.3.

IMPROVE can also be regarded as a tool integration project with a broad understanding of “integration” (cf. part I of this book). This section is to

summarize

our

findings

on tools

by regarding

four

different

views

on these tools: (a) contributions of IMPROVE to a better support of development processes by using tools, (b) lessons learned from tool construction/integration, (c) how tool construction and modeling of development processes interact, and (d) how application-specific or general our results on tools are. The review is restricted to the academic viewpoint. The industrial review for tools – but also for other perspectives – is given in Sect. 8.3.

This short section presents some thoughts on the relevance and the impact of the research work in IMPROVE on industrial practice. Though, the research program has been linked to industrial requirements and has seen a continuous review from industrial colleagues to refocus the research objectives and to assure practical relevance of the IMPROVE research approach, only few concrete results have made it yet into industrial practice. However, the problems addressed in IMPROVE have been receiving significant attention in industry, in particular in the recent past. In this sense, IMPROVE has been addressing a timely and forward-looking fundamental research program, which has come too early to be readily absorbed by industry. However, it will be of significant impact on industrial practice in the future in both, the software as well as the chemical and process industries.

This short section aims at evaluating the academic outcome of the CRC IMPROVE from 1997 up to now and also sketches its further outcome as TC 61. We do this by regarding different perspectives: (a) The number and value of publications, (b) the contribution to international conference activities, and (c) how we laid the basis for or accelerated the career of young scientists. Joint activities together with industry and their consequences are discussed in sections 5.1 and 8.3.