Sustainable Manufacturing pp 239253  Cite as
Optimisation Methods in Sustainable Manufacturing
Abstract
Sustainable manufacturing is driven by the insight that the focus on the economic dimension in current businesses and lifestyles has to be broadened to cover all three pillars of sustainability: economic development, social development, and environmental protection.
1 Introduction
Sustainable manufacturing is driven by the insight that the focus on the economic dimension in current businesses and lifestyles has to be broadened to cover all three pillars of sustainability: economic development, social development, and environmental protection. In this chapter, we present two stateoftheart approaches of mathematical optimisation and how they can be used to solve problems in sustainable manufacturing.
The multicriteria perspective considers areas of sustainability as independent functions that are to be optimised however with divergent objectives simultaneously. Accordingly, computed outcomes that cannot be improved upon (on at least one objective without getting worse at another) are considered to be superior to outcomes that can be improved upon. A decision maker will only be interested in the first set of outcomes in order to be able to form an educated opinion with respect to his/her sustainability goal.
The system dynamics perspective on the other hand focuses on the timedependent (or dynamic) aspects of systems that are influenced by sustainable manufacturing practices. If, for instance, a production technology was identified that cannot be improved in either of the sustainability dimensions, the question then arises as to how this technology can be used in an optimal way using only limited resources. How can the impact on society and economy be steered in the direction of allowing the technology to be as beneficial as possible?
2 Multicriteria Optimisation
Mathematical optimisation and mathematical programming is concerned with finding good solutions from a set of available alternatives. The abstract nature of mathematical optimisation allows the user to model a wide range of different problems and different objectives using the same theoretical insights and practical tools. Problems in sustainability and sustainable manufacturing have in common that there is not only one objective to be considered but several conflicting ones. This is mathematically reflected by considering several objective functions simultaneously. The set of available alternatives and the structure of the considered objective functions can generally be modelled in different ways. The focus in the following section is put on the wellstudied and fruitful field of linear optimisation involving linear objective functions and linear constraints allowing the user to model as well as to efficiently solve a wide range of quantitative problems.
2.1 Multicriteria Problem Formulation
2.2 Manufacturing and Scheduling
2.3 Solving Multicriteria Optimisation Problems
For the singleobjective case there are several commercial solvers and software packages (CPLEX 2016; Xpress 2016; Gurobi 2016) and noncommercial ones (Achterberg 2009). One could have expected that the exponential growth in computing power and the even larger algorithmic speedups in mixed integer programming during the last decade (Bixby 2002) would automatically lead to multicriteria extensions. But the situation is contrary: none of the available commercial solvers supports multicriteria problems and there are only a few, recently developed noncommercial solvers available: BENSOLVE (Löhne and Weiing 2014) and inner (Csirmaz 2016) handle multicriteria linear programming problems, SYMPHONY (Ladanyi et al. 2016) supports bicriteria mixed integer problems and PolySCIP (Schenker et al. 2016) supports multicriteria linear and integer problems.
3 System Dynamics Optimisation
In this book, many technologies and approaches developed in the context of sustainable manufacturing are discussed. In this section, we will consider the global environment in which these technologies must be disseminated and implemented, in order to realise their positive potential.
The economy, the environment, and the society constitute complex entities and can be seen as finely balanced networks of mutual dependencies. Almost all components influence each other that have either supporting or weakening effects. Such dynamical systems can demonstrate counterintuitive behaviour. However, in order to bring about a change from the conventional production paradigm in the direction of a paradigm of sustainability, it is essential to appreciate the complex interdependencies of the systems involved.
We observe that the transition, i.e., the setup of many value creation modules and networks, constitutes a dynamic process over time that will span several years or decades. During this period, an array of interactions between the stakeholders need to be taken into account. Moreover, the transition does not take place by itself. It will only happen by means of deliberate influence on the system. A bundle of individual measures are necessary in this process.
To this end, the system dynamics (SD) approach provides the appropriate framework. It is an approach for the modelling and simulation of dynamical systems with a long history rooted in the understanding and teaching of dynamical systems in general, as well as in the field of sustainability.
After introducing system dynamics as a tool for simulation, we will formulate optimal control problems based on system dynamics models.
3.1 System Dynamics
In this section, we will introduce system dynamics as a modelling methodology as well as the most important modelling rules and characteristics of system dynamics models.
System dynamics was introduced by Jay Forrester in the 1950s as a method of describing and simulating timedependent effects of complex influence networks with feedback loops (Forrester 1961). Such networks are characterized by nonlinear, often surprising behaviour. In fact, a forecast of their future development, and thus their control, represents a difficult mathematical problem.
One of the strengths of the system dynamics approach lies in its visual representation of complex systems. This visual approach is essential in the system dynamics modelling process, and simplifies access for beginners and users who lack experience with systems of differential equations.

Definition of the modelling goal,

Definition of the system limits,

Definition of the system components,

Definition of the direct relations between system components and the type of causal links (positive or negative),

Design of an influence diagram to summarize components and their relations,

Creation of a system dynamics diagram with stocks for each of the system components as well as flows for each stock,

Assignment of units and valid ranges to the values of stocks and flows,

Definition of the functional relations between stocks and flows,

Introduction of variables to simplify the relations if possible,

Completion of the system dynamics diagram by adding variables and arrows for relations,
The result of this process is a complete system dynamics model. In the next section, we will discuss numerical methods for simulating a system dynamics model as it develops over time.
Although it is possible to find general solutions analytically for some models, this is generally neither possible nor required. A range of numerical simulation techniques exist that provide quickly accurate simulations. One class of such simulation techniques are the RungeKutta schemes (Runge 1895; Kutta 1901) which we will use in this chapter.
3.2 Optimal Control of System Dynamics Models
As we discussed in the previous sections, in its basic form, SD aims at describing and simulating influence networks. This is an important step in pursuit of understanding the mutual dependencies. In addition to obtaining a mere understanding however, what we would like to do is to intervene in the network, bring it to a desired stable state, or get as close as possible to that state.
In system dynamics, the points of the system which can be influenced by a conscious decision of an actor are modeled using the concept of policies.
Policies constitute a basic and important concept of system dynamics modelling. A policy is a function in some variables that describes the rates of flow in a system and hence the dynamic behaviour of the model (Richardson and Pugh 1981). Thus, a policy is a decision rule which specifies how a decisionmaker processes available information from model variables (Sterman 2000). Questions regularly arise concerning whether a given policy can be improved, or even what a “good” policy “actually constitutes or entails. In this context, the need for efficient computational methods for policy analysis as well as policy improvement and design has been recognized in system dynamics, see, e.g., Yücel and Barlas (2011), Keloharju and Wolstenholme (1988), and is an active field of research.
When developing a simulation model, the modelling step of “policy formulation and evaluation” also compares the performance of two or more candidate policies (Sterman 2000). When two simulations with different policies lead to different system behaviors, one has to evaluate which of the two simulations is more suitable or “better” for a given model purpose. To answer this question, one needs to define an objective function so that the higher the value of the objective function for a given simulation, the more favorable or “better” the policy (Dangereld and Roberts 1996). Once an objective function is defined, several approaches to computeraided policy improvement are at one’s disposal.
Direct parameter policy design starts with the definition of an analytic, parametrized, and usually nonlinear policy function (Keloharju and Wolstenholme 1989). The parameters of this function are set to starting values, and for each parameter, a range of valid values is defined. These parameters constitute then the free variables of the optimisation problem, i.e., the variables which can be varied freely in pursuit of an optimal solution. Consequently, the goal of the policy improvement is to find a set of parameter values within the given range that improves the value of the objective function. The solution space in this case is reduced by the a priori definition of the shape of the policy function. The solution found by the optimisation algorithm depends strongly on this definition and therefore on the expectations of the modeler. If a software package offers parameter optimisation capabilities, it is usually possible to attempt producing the solution of such direct parameter policy design problems.
Table function policy design is one possible way to generalizing direct parameter policy design, by defining a parametrized table function instead of an analytic function (Keloharju and Wolstenholme 1989). In this case, the modeler has to define the number of data points of the table function and two intervals that define the range of valid values of the data points on the \( x \) and \( y \)axis. This approach removes the modeler’s expectations of the shape of the policy from the optimisation process. However, the possible policies are reduced to the space of the piecewise linear functions with the selected number of points. If the data points are then required to have a predefined distance on the \( y \)axis, the possible solutions are reduced further, but at the same time, the number of parameters and thus the number of free variables decreases. As in the previous case, the goal of the policy improvement is to find parameter values (i.e., data points of the table function), that improve the value of the objective function. A software package that supports table function policy design is found with the Powersim Studio plugin SOPS (Moxnes and Krakenes 2005).
In both cases, the modeler has to define the functional dependencies of the policy function. This choice is closely related to the concept of bounded rationality (MoreCroft 1985; Simon 1984) models.
A policy function, i.e., a decision rule, is a model about what information cues an actor employs in order to make decisions in a given system. If this actor has only a limited view of the system, then the policy will only depend on the variables and information that are available to this particular actor (Sterman 2000). An improved policy will enable this actor to make better decisions based on the limited information available to him/her. Recent work has focused on improving policies for such actors, using, for instance, coevolutionary analysis (Liu et al. 2012).
In this paper, we will consider a different kind of actor. Our actor has a global view of the model, i.e., he or she has information on all the state variables at all times within the simulation time horizon.
Modeling the policy of an actor with such a comprehensive level of awareness with the application of conventional approaches to policy analysis constitutes a difficult endeavor. One option would be to define a table function for each state, that depends only on that state. A mixed policy function that depends on all states, can then be defined as a sum of these functions (Keloharju and Wolstenholme 1989).
One conventional approach to System Dynamics optimisation is based on “optimisation by repeated simulation” (Liu et al. 2012). This has the advantage, that any model which can be simulated, can also be optimized, since there are no requirements on the properties of the model equations. However, approaches using repeated simulation suffer from the “curse of dimensionality” Bellman (2003) dynamic, where the significant dimension is that of the space of free variables. An additional free variable adds a dimension to the optimisation algorithm’s search space. Solving optimisation problems with a large number of free variables therefore quickly becomes impractical. As a consequence, the degrees of freedom in a mixed policy function situation, are limited from a practical perspective, in the case of an optimisation of the policy by repeated simulation being attempted.
We present a different approach and in so doing, directly optimize the values of the policy function. This is equivalent to defining the policy as a timedependent table function with one data point for each time step of the time horizon. In the context of physical systems, this kind of problem is known as an “optimal control problem” Betts (2011). With this approach no assumptions on the properties of the policy function are made a priori. It is only necessary to select the “free variables”. In a conventional approach, these “free variables” would contain the values of the policy functions. For each of these variables, a range of valid values must be defined. It is then the task of the optimisation process, to find the optimal value for each free variable at each time.
In order to solve such a problem, we differentiate between two approaches:
3.2.1 Local Approach
In the local approach, the goal is to find a locally optimal solution. Local optimality means, that in a small neighborhood around the given solution, there is no solution with a better objective value. For this approach, standard methods exist for dynamical systems, which reliably deliver local solutions for small and moderately sized problems. The task at hand is to reformulate and adapt a system dynamics model, so that these methods can be used. Work on the local optimisation of system dynamics models can be found for instance in Vierhaus et al. (2014). In this chapter, we will focus only on the global approach.
3.2.2 Global Approach
In the global approach, the goal is to find a solution, and in addition to prove its global optimality. This means that no feasible solutions of the problem with a better objective function value exist. Hence, the global solution approach has two steps: Find an optimal solution and prove that no better solution exists.
Both of these approaches can prove successful using techniques from mathematical optimisation.
In the next section, we will show how modern optimisation techniques can be used in the global approach to system dynamics optimisation. The basis is the formulation of an optimisation problem, based on the control problem introduced in Sect. 3.2. As mentioned before, the simulation of a system dynamics model using numerical methods is wellestablished. This simulation is based on a timediscretisation of the model, which we will also use for our optimisation problems.
This system now has the standard form of an optimisation problem, similar to the one introduced in (1). In contrast to (1), we now only have a single objective function. On the other hand, we have nonlinear equality constraints in place of linear inequality constraints.
3.3 MINLP Approach
After the discretization of the system dynamics optimisation problem, it is possible to attempt to solve it with existing solvers. Since we are interested in global solutions, the algorithm used should be able to provide a certificate of global optimality. One group of solvers that can provide this certificate are the branchandcut solvers that were introduced in Sect. 2.3 This approach has been successfully applied in the solution of Mixed Integer Linear Programs as well as MINLPs from a range of applications [for example, see Defterli et al. (2011), Borndörfer et al. (2013), Humpola and Fügenschuh (2013)]. Solving a control problem derived from a discretised dynamical system with a standard branchandcut solver is, however, in many cases unsuccessful, since the solver does not take into account the special structure of the MINLP that arises from the discretization, and from the handling of nonsmooth functions via integer variables. Without considering this structure, even finding a single feasible solution can exceed a reasonable time budget of several hours or even days.
In the remainder of this section, we will present the concept of a tailored solver for system dynamics optimisation problems. Like PolySCIP, this concept has been implemented in the framework of the modern MINLP solver SCIP and results can be found in Fügenschuh and Vierhaus (2013a, b), Vierhaus et al. (2014), Fügenschuh et al. (2013). A diagram describing the improved solution process is shown in Fig. 6.
3.3.1 Transcription
3.3.2 Optimisation Based Reachability Analysis
To improve on the dual side of the algorithm, an Optimisation Based Reachability Analysis (OBRT) is performed for every problem. This analysis computes bounds for the possible states of the system using the dynamic behaviour and the initial values \( x_{0} \) as input.
3.3.3 Primal Heuristic
In the interest of producing quickly feasible solutions, we implemented a simple heuristic that reduces the control problem to a simulation problem by fixing the control variables to their lower (or in a second run upper) bounds. If there are no path constraints, this process will always yield a feasible solution.
3.3.4 Bound Propagation Based on Differential Inequalities
To improve the bounds within the branchandcut process, we compute differential inequalities as outlined in Scott and Barton (2013). This involves the solution of an auxiliary simulation problem using the expressions for \( \dot{x} \) derived in the reading of the problem.
3.3.5 System Dynamics SCIP
The concepts mentioned above have been implemented as the solver System Dynamics SCIP (SDSCIP). Like polyscip, SDSCIP is an extension of the modern MINLP solver SCIP and is publicly available (Füegenschuh and Vierhaus 2013a, b).
4 Conclusion
This chapter introduced the framework of multicriteria optimization and system dynamics optimisation together with different modelling techniques. It showed that mathematical optimisation is a useful tool for modelling a wide variety of problems from the sustainability context. The two solvers presented PolySCIP (Schenker et al. 2016) and SDSCIP (Fuegenschuh and Vierhaus 2013a, b) were specifically developed with applications from sustainability in mind. They can be used as decision support instruments for a wide range of problems, from scheduling, manufacturing and production to planning subsidies and taxes and exploring dynamical pathways into the future. Both tools are publicly available and present an opportunity for the sustainability community to benefit from recent advances in mathematical optimisation.
References
 Achterberg, T. 2009. Scip: Solving constraint integer programs. Mathematical Programming Computation 1(1): 1–41. http://mpc.zib.de/index.php/MPC/article/view/4.
 Bellman, R. 2003. Dynamic programming. Dover Books on Computer Science Series. Dover Publications.Google Scholar
 Betts, J.T. 2010. Practical methods for optimal control using nonlinear programming. Advances in design and control. Philadelphia, PA: Society for Industrial and Applied Mathematics.zbMATHGoogle Scholar
 Bixby, R.E. 2002. Solving realworld linear programs: A decade and more of progress. Operations Research 50: 3–15.MathSciNetCrossRefzbMATHGoogle Scholar
 Borndörfer, R., A. Fügenschuh, T. Klug, T. Schang, T. Schlechte, and H. Schülldorf. 2013. The freight train routing problem. Technical report, ZIB Technical Report ZR1336.Google Scholar
 CPLEX. 2016. IBM ILOG CPLEX Version 12.6.3. www03.ibm.com/software/products/en/ibmilogcpleoptistud.
 Csirmaz, L. 2016. Inner. https://github.com/csirmaz/inner.
 Dangerfield, B., and C. Roberts. 1996. An overview of strategy and tactics in system dynamics optimization. Journal of the Operational Research Society 47: 405–423.CrossRefGoogle Scholar
 Dantzig, G. 1963. Linear programming and extensions. Rand Corporation Research Study. Princeton University Press.Google Scholar
 Defterli, O., A. Fügenschuh, and G.W. Weber. 2011. Modern tools for the timediscrete dynamics and optimization of geneenvironment networks. Communications in Nonlinear Science and Numerical Simulation 16(12): 4768–4779.MathSciNetCrossRefzbMATHGoogle Scholar
 Ehrgott, M. 2005. Multicriteria optimization, 2nd ed. Berlin: Springer.zbMATHGoogle Scholar
 Fügenschuh, A., and I. Vierhaus. 2013. A global approach to the optimal control of system dynamics models.Google Scholar
 Forrester, J.W. 1961. Industrial dynamics. Waltham, MA: Pegasus Communications.Google Scholar
 Fügenschuh, A., S.N. Grösser, and I. Vierhaus. 2013. A global approach to the control of an industry structure system dynamics model. Technical Report 1367, ZIB, Takustr.7, 14195 Berlin.Google Scholar
 Fügenschuh, A., and I. Vierhaus. 2013. A global approach to the optimal control of system dynamics models. Technical report, ZIB Technical Report ZR1328.Google Scholar
 Gurobi. 2016. GUROBI Optimization Version 6.5. www.gurobi.com.
 Humpola, J., and A. Fügenschuh. 2013. A unified view on relaxations for a nonlinear network flow problem. Technical report, ZIB Technical Report ZR1331.Google Scholar
 Keloharju, R., and E. Wolstenholme. 1988. The basic concepts of system dynamics optimization. Systemic Practice and Action Research 1(1): 65–86.Google Scholar
 Keloharju, R., and E. Wolstenholme. 1989. A case study in system dynamics optimization. Journal of the Operational Research Society 40(3): 221–230.CrossRefGoogle Scholar
 Koch, T. 2004. Rapid mathematical prototyping. Ph.D. thesis, Technische Universität, Berlin.Google Scholar
 Kutta, W. 1901. Beitrag zur näherungsweisen Integration totaler Differentialgleichungen. B.G Teubner.Google Scholar
 Ladanyi, L., T. Ralphs, M. Guzelsoy, and A. Mahajan. 2016. SYMPHONY 5.6.14. https://projects.coinor.org/SYMPHONY.
 Liu, H., E. Howley, and J. Duggan. 2012. Coevolutionary analysis: A policy exploration method for system dynamics models. System Dynamics Review 28(4): 361–369.CrossRefGoogle Scholar
 Löhne, A., and B. Weißing. 2014. Bensolve 2. http://www.bensolve.org.
 Morecroft, J.D.W. 1985. Rationality in the analysis of behavioral simulation models. Management Science 31(7): 900–916.CrossRefGoogle Scholar
 Moxnes, E., and A. Krakenes. 2005. SOPS—a tool to find optimal policies in stochastic dynamic systems. In Proceedings of the 23rd International Conference of the System Dynamics Society, ed. by J.D. Sterman, N.P. Repenning, R.S. Langer, J.I. Rowe, and J.M. Yanni.Google Scholar
 MPS Format. 2016. Mps format—Wikipedia, the free encyclopedia. Accessed 28 July 2016.Google Scholar
 Richardson, G.P., and A.L. Pugh. 1981. Introduction to system dynamics modeling with dynamo. Cambridge, MA: MIT Press.Google Scholar
 Runge, C. 1895. Ueber die numerische Auflösung von Differentialgleichungen. Mathematische Annalen 46: 167–78.MathSciNetCrossRefzbMATHGoogle Scholar
 Schenker, S., R. Borndörfer, M. Skutella, and T. Strunk. 2016. PolySCIP. In Mathematical Software –ICMS 2016, 5th International Congress, Proceedings, ed. by G.M. Greuel, T. Koch, P. Paule, and A. Sommese, vol 9725 of LNCS, Berlin, Germany. Springer. http://polyscip.zib.de.
 Scott, J.K., and P.I. Barton. 2013. Bounds on the reachable sets of nonlinear control systems. Automatica 49(1): 93–100.MathSciNetCrossRefzbMATHGoogle Scholar
 Simon, H.A. et al. (1984). Models of bounded rationality, volume 1: economic analysis and public policy. MIT Press Books 1.Google Scholar
 Sterman, J.D. 2000. Business dynamics—systems thinking and modeling for a complex world. Boston: Irwing McGrawHill.Google Scholar
 Vierhaus, I., A. Fügenschuh, R.L. Gottwald, and S. Groesser. 2014. Modern nonlinear optimization techniques for an optimal control of system dynamics models. In Proceedings of The 32nd International System Dynamics Conference.Google Scholar
 Xpress. 2016. FICO Xpress Optimization Suite Version 7.9. www.fico.com/en/products/ficoxpressoptimizationsuite.
 Yücel, G., and Y. Barlas. 2011. Automated parameter specification in dynamic feedback models based on behavior pattern features. System Dynamics Review 27(2): 195–215.CrossRefGoogle Scholar
Copyright information
Open Access This chapter is licensed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.
The images or other third party material in this chapter are included in the book’s Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the book’s Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder.