A recent VCI-Prognos-study funded by the “Ver-band der Chemischen Industrie e.V.”, a Germanassociation of the chemical industry, describes thefuture of the chemical industry until the year 2030(VCI-Prognos-Study, 2012). The forecast shows agrowing chemical market in which industrial coun-tries can only compete due to their highly integra-ted and efficient production networks, called “Ver-bund”. While the German chemical industry redu-ced energy consumption by one fifth compared to1990, the production volume increased nearly 60%in the same period due to process optimization andeffects of integrated production networks. Accor-ding to the study, increasing costs of raw materi-al and energy will force the German chemical indus-try to further increase their resource efficiency inorder to continue to be competitive on the futureglobal market. It is expected that the German che-mical industry will increase its energy consumpti-on by only 8 % while the production value willincrease by 40%. This efficiency increase is also dueto a change in the product mix: While the volumeof high-end chemicals will increase, resource inten-se products will grow slower than the overall mar-ket. But, how to adapt the production network tomeet future demands? This article gives a short
overview of extended material flow analysis anddemonstrates in a simplified case study its appli-cation to define proper mid to long-term develop-ment strategies for integrated production net-works.
2 Integrated production networks
An integrated production network contains aweb of production plants cross-linked by materialand energy flows (Brudermueller, 2001, Brudermu-eller and Langguth, 2001, Viere, 2009). The usageof co-products of single plants as raw material forother plants or as a recycle flow to upstream pro-duction units can lead to high resource producti-vity and it reduces the amount of waste and emis-sions. Also, energy flows between the productionunits at different enthalpy levels reduce the over-all energy consumption of the network leading tohigh energy efficiency. Organized in an optimizedmanner, integrated production networks are costeffective and, therefore, deliver competitive advan-tages. However, the complexity of large networksmakes it difficult to predict and navigate throughvolatile markets and to define development stra-tegies to meet future market demands.
There is a variety of literature (e.g., Grossmann,2005, Proud, 2007, Duggan, 2012) about the ope-
Practitioner’s SectionDevelopment of integrated production net-works using extended material flow analysis
Marco Auer*
Integrated production networks can be very efficient in using raw materials andenergies resulting in optimized cost structures. In particular in the chemical indus-try integrated production networks gain competitive advantage. However, thecomplexity of large networks makes it difficult to predict and navigate throughvolatile markets and to define development strategies to meet future marketdemands. But, by using extended material flow analysis effects of changes in pro-duction networks can be predicted. Starting with material flow analysis and exten-ding with cost and investment analyses while using scenario techniques, an opti-mized configuration of the network can be identified. Therefore, extended mate-rial flow analysis supports the strategic development of the integrated producti-on network.
Development of integrated production networks using extendedmaterial flow analysis
35
rational planning of production units focusing onshort and mid-term horizon to optimize man powerand machine uptime, reducing change over time,maximizing productivity and minimizing logistictraffics. Typically tools and algorithms are used forproduction scheduling considering production ona high detail level but in a relatively small networksubarea, e.g., one production line or one chemicalplant.
Tactical and strategical planning have a widerprojection horizon and, in contrast to operationalplanning, they have to deal with more options anduncertainties. The plans are based on estimatedmarket demands on different planning horizon of2, 5, 10 or even more years. They should ensure tofollow the business strategy by developing the pro-duction network accordingly. In order to sustainfuture profitability and efficiency in the networkthe following questions come up frequently:
Does the existing integrated production net-work serve future market demands and price structures? If not, what bottlenecks must be opened, how should the network be adapted?How can the productivity of the overall net-work be increased?Is the production network able to follow thestrategy and vision of the company?Extended material flow analysis contributes to
answering these questions. Its professional deploy-ment supports the development of an efficient,economic, and ecological production network.During the analysis phase it
gives a task-oriented and transparent view ofcomplex material and energy flows and coststructures,detects constraints and bottlenecks,reveals dependancies, correlations, and inhe-rent product mixes,identifies profitable operating corridors of theexisting network,evaluates the product portfolio, anddefines meaningful key figures.In further investigations using scenario techni-
que extended material flow analysis depicts impro-vements with clear targets and possible solutions
to debottleneck and optimize production capa-bilities,to adapt the product portfolio correspondingto future demands,to increase productivity and resource efficien-cy, e.g., by cycle flows of intermediates, fully uti-lized coupled products, waste minimization,optimal energy utilization, to meet safety and environmental requirements,e.g., minimize inter-site transportation of hazar-dous materials, combined exhaust, sewage andwaste treatment,
to support business risk assessments, e.g., shut-down of plants, limited raw material availabi-lity, big drop in prices or demands (Alberti, 2001).Overall, extended material flow analysis prepa-
res and supports strategic decisions in order todevelop further the production network of a com-pany. In most cases the investigation results in clearproject definitions with task, target, and potenti-al benefit. In combination with estimated invest-ment costs and expected sales numbers the pay-back, return on investment (ROI) and internal rateof return (IRR) can be calculated.
3 Extended material flow analysis
This approach is based on material flow analy-sis described in Moeller et al., 1995, and Brunner etal., 2003. A converged material and energy networkmodel provides the basis for the cost calculation.The subsequent analysis of the model leads toadjustments in the model. It is an iterative approachto optimize the network according to the target.
Several software applications are available onthe market for material flow analysis including costanalysis. Brunner and Rechberger (Brunner et al.,2003) compared several tools with the similar exam-ple including GaBi (http://www.gabi-software.com)and Umberto (http://www.umberto.de/en/). Bothtools were further developed since their compari-son. Umberto was recently recommended by Gart-ner Inc. in their “Cool Vendors for Green IT and Sus-tainability 2012” report (Gartner Inc., 2012). For theanalysis discussed in this article Umberto was usedfor material flow analysis and Sankey diagramsand Microsoft Excel for additional analyses, tables,and diagrams.
3 . 1 Modeling of the network & boundaryconditions, initial flows
Models in material flow simulation contain tran-sitions, arrows, and places. In transitions the trans-formation of material and energy flows are calcu-lated. Places represent nodes in the model whichmay be just connections, stocks for material andenergy, or boundary conditions at system borders(input and output). The arrows define possible con-nections between places and transitions. The flowdirection and amount will be calculated by the algo-rithm. Depending on the investigation, the modelsize and its boundary conditions may represent asubarea of a production site only or the full pro-duction network with all different sites.
The level of detail of the model is related to thetarget of the investigation (Bode et al., 2011). If, e.g.,the task is to analyze and optimize heat exchangernetworks the details of all relevant operation units
and their material and energy flows have to bemodeled (e.g., reactors, vaporizers, distillationcolumns, condensers, dryers). In contrast, for ana-lysis of inter-site transportation, the modeling onplant level is usually sufficient, i.e., modeling ofmajor operation units or on plant level with mainflows between the units or plants, respectively. Itis strongly recommended not to overload modelssince a very high level of detail increases the com-plexity of the model significantly, that leading toconvergence issues and incomprehensible results.
Typically, material flow simulation uses theapproach of a Petri net to calculate the network.That is a sequential algorithm in which every tran-sition calculates separately all possible equationsdepending on new inward or outward directedflows. To start the simulation initial flows have tobe defined. The initial flows may represent a mar-ket pull or a push from the raw material side at theboundary of the system or defined flows withinthe network. Also a combination of several flowscan start the calculation. In complex networks,several initial flows have to be defined in order toget the full model calculated. The algorithm startsat the transitions where the initial flows are queued.If all transitions of the network are calculated ordo not change in a following run the simulationends.
After successful calculation of the material andenergy network a post algorithm starts the costcalculation with additional cost information, e.g.,fix and variable costs, energy costs, allocation rules.In production networks of the chemical industrythere are usually several units with by-products,i.e., more than one product is produced simultane-ously in these units. The allocation of the joint coststo all products of this unit may have huge impacton the economic efficiency of final products and,therefore, must be modeled correctly (Langguthand Brudermueller, 2001, Fandel et al., 2009). As agood example, Bode et al. describe the influenceof allocation rules on the economic evaluation ofdifferent process options (Bode et al., 2011).
3 . 2 Analysis
Typically the analysis of the model will be car-ried out using several methods in combinationdepending on the type of investigation. Methodsemployed commonly include the following:
Balance sheets to list flows across a defined sys-tem boundary. The system can be the overallnetwork, subareas of the network, or single units(Moeller and Rolf, 1995). Multiple variants arepossible, e.g., material or energy flows and theirassigned values, grouped by material, listed byarrows, etc..
Pareto, ABC, or portfolio analysis (Daenzer et at.,1994, Lunau et al., 2008) to cluster raw materi-al, intermediate, and product flows and to eva-luate the portfolio.Flow and cost analysis via Sankey diagrams tovisualize product or product group flows andtheir cost structures and to detect constraints,dependancies, and bottlenecks (Schmidt,2008a&b).
3 . 3 Optimization
After detailed analysis of the current model, thenext step is to optimize or further investigate thenetwork according to the task. Here, scenario tech-nique is often used in material flow analysis tomodify the network and benchmark the results orto identify cause-and-effect chains. The modifica-tions in the network vary from adjusted initial flowsto structural changes by new flows, new producti-on units, or new technology with different selecti-vity modeled in transitions. For risk assessments,drastic changes should be modeled, e.g., shut downof internal power plants or important operationunits, shortage of raw material. After the modifi-cations, the new scenario has to be recalculatedand analyzed.
By comparing different scenarios, sensitivityand regression analysis may help to detect rootcauses for constraints and dependancies. Bench-marks of the scenarios are used to evaluate the dif-ferent modifications revealing favorable networkconfigurations or most economic modes of opera-tion. Once identified, the way from the existingnetwork to the desired configuration can also bedefined by using extended material flow analysis.
A very positive side effect of using extendedmaterial flow analysis for development of integra-ted production networks is the identification ofclear tasks and targets for following projects whichare defined by the results of the analysis. With thesefront-end loads the project teams have a clear pictu-re of what has to be implemented.
4 Case study Mueller- Rochow Direct Syn-thesis of methylchlorosilanes
The following case study is a brief and simpli-fied example to demonstrate how extended mate-rial flow analysis can support strategic decisionsin the chemical industry with the aim to adapt theintegrated production network to future marketdemands. According to data confidentiality, thecase is based on process descriptions published in“Silicones & Industry: A compendium for practicaluse, instruction and reference” by Andreas Toma-nek (Tomanek, 1992).
Development of integrated production networks using extendedmaterial flow analysis
The Mueller-Rochow Direct Synthesis is a cop-per-catalyzed reaction of chloromethane with sili-con in a fluidized bed reactor. Almost all methyl-chlorosilanes are produced by this direct synthe-sis. Methylchlorosilanes are the raw materials forsilicone polymers and oils, resins and organo-functional silanes (Tomanek, 1992).
Modern fluidized bed reactors have approxima-tely 40,000t annual capacity of raw silane, a mix-ture of different methylchlorosilanes. The usualcomposition according to Tomanek (Tomanek, 1992)is shown in Table 1.
4. 1 Base scenario: 85% of reactor capacity
Usually these large scale fluidized bed reactorsare implemented in highly integrated productionnetworks with voluminous cycle streams of inter-mediates. In figure 1, a simplified production net-work is shown. The transitions are shown as rectan-gles with an identifier starting with the letter “T”,e.g., the transition T1 is the Mueller-Rochow DirectSynthesis. The places are shown as circles wheregreen circles with a secant on the left side are inputplaces while red circles with secants on the rightare output places. Input places represent the sys-tem boundary upstream, where materials enterthe system. Output places represent the outlet ofthe system. In this model, the output places P3, P8and P9 represent further downstream productionlines not discussed in this case. For all internal pla-ces in this simplified model, stocks are not allowed.The arrows are shown in different colors and linewidth. The color defines the material or materialgroup and the width is proportional to the flowquantity. This type of flow diagram is known asSankey diagram (Schmidt, 2008a&b). The numbersclose to the arrows are the calculated mass flows
in tons. All numbers discussed are rounded num-bers.
In this scenario, 1,541t of raw silicon are reactingin the Mueller-Rochow Direct Synthesis (transiti-on T1) with 5,681t of chloromethane (blue cycle stre-am) to a mixture of raw silanes (green arrow, 7,102t).This should represent 85% of total reactor capaci-ty of a certain period. 120t of waste out of the Muel-ler-Rochow Direct Synthesis leave the modelthrough the output place P2. In the distillation T2,the raw silane mixture will be separated and high-boiling methylchlorosilanes (approximately 2% ofthe mixture) will be mostly recycled with the aidof HCl and an amine catalyst to dimethyldichloro-silane. The separated products are listed in Table2.
From the place P5, the outlet of the distillation,the different silanes are distributed to severaldownstream transitions:
In the Methanolysis 1 and 2 (transitions T4and T5) dimethyldichlorosilane and trimethylchlo-rosilane react with methanol forming dimethylsi-landiol and trimethylsilanol, respectively, precur-sors for polycondensation to silicone polymers andoils. The by-product chloromethane can be recy-cled as reactant for the Mueller-Rochow Direct Syn-thesis. This very important recycle stream is repre-sented by the blue arrow from P7 over T13 to P6and finally to T1.
Methyltrichlorosilane out of the distillationreacts with methanol to methyldichloromethox-ysilane and HCl in the Alkoxylation (T8). Methyldi-chloromethoxysilane is a precursor for resins. HClcan be recycled in T12 with methanol to form chlo-romethane for the Mueller-Rochow Direct Synthe-sis and as reactant to convert high-boiling methyl-chlorosilanes to dimethyldichlorosilane in the dis-tillation as described above.
Figure 1 Base scenario of a simplified production network.
Silane wt-% abs. flow
Dimethyldichlorosilane 83% 5899 t
Trimethylchlorosilane 3% 213 t
Methyltrichlorosilane 10% 710 t
Methylhydrogendichloro-silane
4% 284 t
Table 2 List of intermediates after distillation.
P1: RawMaterials
1541 t
T1: Müller-RochowDirectSynthesis
P2: Waste
P3: Poly-mers, Oils
5681 t
P6:CH3Cl
733 t
149 t
120 t
7102 t
234 t
T13: CH3ClRecycle
5447 t
4 t
T2:Distillation
T12: HClRecycle64 t
4714 t
2929 t
P5
5899 t
T4: Metha-nolysis
213 t
63 tT5: Methanolysis 2
710 t
152 t
169 tT8: Alkoxylation
284 t
T10: Hydro-silytion
4212 t
4615 t
P7: CH3Cl
177 t
689 t
173 t
P10: HCL
348 t
P8: Resins
P9: organo-functionalSilanes
In the Hydrosilylation (T10) methylhydrogen-dichlorosilane reacts with acetylene to form vinyl-methyldichlorosilane as example of numerous pos-sible hydrosilylations to generate precursors fordifferent organo-functional silanes.
A balance sheet over the whole network of thisscenario shows that 1,541t of silicon are needed toproduce 4,212t of dimethylsilandiol, 177t of trime-thylsilanol, 689t of methyldichloromethoxysilaneand 348t of vinylmethyldichlorosilane. Only 733t ofchloromethane are from external resources while5,681t are used in the same period in the Mueller-Rochow Direct Synthesis. About 4,615t or 81% ofthe required chloromethane are a by-product ofMethanolysis 1 in this scenario.
In the current scenario the network producesan inherent product mix in which the output of allintermediates is fixed to a certain ratio defined bythe selectivity of the Mueller-Rochow Direct Syn-thesis and its throughput. The market prices of rawmaterials and energies and the costs allocationsin the Mueller-Rochow Direct Synthesis and thedistillation define the internal prices of the vario-us intermediates and finally the price of the endproducts of all products downstream.
4. 2 Target: Adapt production network tochanged market demands
In this case study, a market survey reveals hig-her demand on organo-functional silanes whilethe rest of the market remains stable. This leadsto a higher demand on methylhydrogendichloro-silane as a precursor for different organo-functio-nal silanes.
In the following sections, several scenarios willbe investigated using extended material flow ana-lysis to find the best adaption of the integratedproduction network in order to meet the futuremarket demands. In a first step, several assumpti-ons are made to simplify the study:
enough raw materials and utilities are availa-bleno bottlenecks in production units, cycle stre-ams, and downstream processessufficient separating efficiency of distillationsand other separation units
4. 3 Scenario A: Full reactor capacity
The overall throughput of the Mueller-RochowDirect Synthesis will be increased from 85% to 100%(1.2 times) in this scenario A. In figure 2, the Sankeydiagram of this scenario is shown. As expected, allflows are increased by 1.2 times. A balance sheet ofthis scenario will show input flows of 1,813t of rawsilicon, 3,874t of methanol and 863t of chlorome-
thane to produce 4,956t of dimethylsilandiol, 208tof trimethylsilanol, 811t of methyldichloromethox-ysilane and 410t of vinylmethyldichlorosilane.
A comparison with the base scenario disclosesthat an increase of only 62t precursors of organo-functional silanes is attended by a 744t increase ofdimethylsilandiol, 31t increase of trimethylsilanoland 122t more methyldichloromethoxysilane. Sincethe market situation will not change for siliconepolymers, oils, and resins these are undesired over-capacities possibly leading to price drops and lowercontribution margin on these final products.
4. 4 Scenario B: Full reactor capacity andadapted selectivity
In order to increase the output of methylhydro-gendichlorosilane without significant change ofthe output of all other methylchlorosilanes the rawsilane mixture must be varied. In an example publis-hed by Tomanek (Tomanek, 1992) the influence ofconcentrations of lead on the selectivity of theMueller-Rochow Direct Synthesis is described (seetable 3). The selectivity of dimethyldichlorosilanedrops significantly with more than 50ppm leadwhereas the selectivity of methyltrichlorosilaneincreases slightly and that of methylhydrogendi-chlorosilane heavily with higher lead concentrati-on. According to Tomanek (Tomanek, 1992), leadhas no influence on the selectivity of trimethylchlo-rosilane but results in lower Si conversion whicheffects on a different operation of the unit whichis not considered in this study.
In this scenario, at 100% throughput of the Muel-ler-Rochow Direct Synthesis with adapted selecti-vity 4,995t of dimethyldichlorosilane are produced,15% less than in the base scenario with only 85%throughput. Still 1.2x of trimethylchlorosilane isproduced according to the higher throughput sincethe lead concentration has no effect on its selecti-vity. Significantly more amounts of methyltrichlo-rosilane and methylhydrogendichlorosilane areproduced compared with the base scenario, 1,663t(2.3x) and 1,414t (5.0x), respectively. Also, the wasteflow increases by a factor of 3.7 related to the basescenario. The Sankey diagram of this scenario isshown in figure 3.
In comparison with the base scenario (see figu-re 1) it shows also the changes in the precursorflows: 645t less dimethylsilandiol and 30t more tri-methylsilanol are produced, the precursors for themain product group of Silicone polymers and oils,while 925t more methyldichloromethoxysilane isproduced, the precursor for resins. The output ofvinylmethyldichlorosilane, the target precursor forthis case study, increases by 1,386t from 348t to1,734t.
Figure 2 Sankey diagram of scenario A with 100% throughput.
Silane < 5 ppm Pb ≥50 ppm Pb
Dimethyldichlorosilane 83% 60%
Trimethylchlorosilane 3% 3%
Methyltrichlorosilane 10% 20%
Methylhydrogendichloro-silane
4% 17%
Table 3 Influence of concentrations of lead on the selectivity of the Mueller-Rochow Direct Synthesis (Tomanek, 1992).
P1: RawMaterials
1813 t
T1: Müller-RochowDirectSynthesis
P2: Waste
P3: Poly-mers, Oils
6684 t
P6:CH3Cl
863 t
175 t
142 t
8355 t
275 t
T13: CH3ClRecycle
6408 t
5 t
T2:Distillation
T12: HClRecycle76 t
5546 t
3446 t
P5
6939 t
T4: Metha-nolysis
251 t
74 tT5: Methanolysis 2
835 t
179 t
199 tT8: Alkoxylation
334 t
T10: Hydro-silytion
4956 t
5429 t
P7: CH3Cl
208 t
811 t
204 t
P10: HCL
410 t
P8: Resins
P9: organo-functionalSilanes
116 t
According to stoichiometric, the ratio of feedstock to the reactor changes from 6,684t chloro-methane (3.69 : 1) in the scenario A to 6,944t chlo-romethane (3.83 : 1) in the scenario B to a constantfeed of 1,813t silicon. Since less dimethyldichloro-silane is produced in this scenario, less chlorome-thane is recycled (4,024t) and accordingly more hasto be compensated by external sources (2,365t ver-sus 863t in scenario A). Depending on the marketprices of raw silicon and chloromethane, costs forwaste disposal and the allocation of costs at theMueller-Rochow Direct Synthesis and the distilla-tion these changes will have a substantial effecton the conversion costs of intermediates and finalproducts.
4. 5 Scenario C: Full reactor capacity at adap-ted selectivity and converter
In scenario B the output of the main precursorfor silicone polymers and oils, dimethylsilandiol, is645t less than in the base scenario. Moreover, thereis more methyldichloromethoxysilane than nee-ded for resins. Tomanek described a process to rear-range trimethylchlorosilane and methyltrichloro-silane to dimethyldichlorosilane in presence of alu-minum chloride (Tomanek, 1992). The networkshown in figure 4 is extended by a converter (T3)undergoing this rearrangement. As initial flow theinput to the converter is set to consume the fullamount of trimethylchlorosilane out of the distil-lation in order to compensate the discrepancy ofthe throughput of dimethylsilandiol in scenario Bto the base scenario as much as possible.
The mass flow of dimethyldichlorosilane to theMethanolysis 1 is due to the converter 95% of theflow in the base scenario leading to 3,990t of dime-thylsilandiol for further processing to silicone poly-mers and oils. Since the full amount of trimethyl-chlorosilane and 343t methyltrichlorosilane are uti-lized by the converter, no trimethylchlorosilane rea-ches the Methanolysis 2 leading to zerotrimethylsilanol. In this case, trimethylsilanol hasto be purchased from external resources since spe-cific mixtures of dimethylsilandiol and trimethyl-silanol are needed to produce silicone polymersand oils. In addition, the reduction of methyltri-chlorosilane feed to the Alkoxylation reduces theamount of methyldichloromethoxysilane by 333trelated to scenario B without converter but still592t more than in the base scenario. The mass flowof methylhydrogendichlorosilane is not influencedby the additional process of the converter. There-fore, the amount of vinylmethyldichlorosilane isequal to scenario B and 5.0 times higher than inthe base scenario.
The higher amount of chloromethane recycled
from the Methanolysis leads to 10% less externalsupply of chloromethane to feed the Mueller-Rochow Direct Synthesis. Depending on raw mate-rial prices, costs allocation rules, and conversioncosts of the converter unit, the reduction of exter-nal supply of chloromethane may counterbalancethe higher cost of the dimethyldichlorosilane dueto the additional converter unit.
4. 6 Benchmark of scenarios
The target of this case study is to find the bestadaption of the integrated production network tomeet the future market demand especially the hig-her demand on precursors of organo-functionalsilanes. After investigating several scenarios theresults must be benchmarked. According to thetarget of the study the mass flows of the precur-sors will be compared first.
In figure 5 the mass flows of the different pre-cursors are shown by the various scenarios. Theoutput of vinylmethyldichlorosilane as the repre-sentative of numerous precursors of organo-functio-nal silanes increased in every scenario comparedto the base scenario. In scenario A with 100%throughput the increase is only by a factor of 1.2while in the scenarios B and C with adapted selecti-vity of the Mueller-Rochow Direct Synthesis theincrease is 5.0 times. There is no difference bet-ween scenario B and C because the added conver-ter unit in the latter has no influence on the out-put of vinylmethyldichlorosilane. Since no specificvolume of the future demand on organo-functio-nal silanes is defined in the case study, the scena-rios cannot be further assessed regarding thethroughput of the organo-functional silanes.
In this case study the future demand on precur-sors for silicone polymers and oils and for resinsremain stable. Only in scenario C the throughputof dimethylsilandiol is nearly equal to the base sce-nario. In scenario A the throughput is 18% or 743thigher while in scenario B the throughput decrea-ses by 15% or 645t. Concerning trimethylsilanol thescenarios A and B are equal with 18% or 31t increa-se. Only in scenario C the output of trimethylsila-nol drops to zero since the converter consumes allof trimethylchlorosilane to compensate the insuf-ficient throughput of dimethylsilandiol. For furt-her investigations the weighting between dime-thylsilandiol and trimethylsilanol must be definedto optimize the converter throughput accordingly.The weighting may be defined by a combinationof market demands and prices and also by the con-tribution margin of final products and the availa-bility of the precursors on the market. Comparingthe outlet of methyldichloromethoxysilane in thedifferent scenarios, the mass flow is always higher
than the market demand referred to the base sce-nario. In scenario A and B additional 122t (1.2x) and925t (2.3x) are produced, respectively. The high sur-plus production of methyldichloromethoxysilanein scenario B is reduced in scenario C by the con-verter unit leading to 592t or 1.9 times higherthroughput relative to the base scenario.
After having discussed output flows of the net-work in the paragraphs above figure 6 shows themajor input flows. In the scenarios only twothroughput levels of the Mueller-Rochow DirectSynthesis are discussed, 85% and 100%. Accordin-gly only two mass flows of raw silicon are conside-red as initial flows, 1,541t in the base scenario and1,813t in all other scenarios.
Chloromethane as the second reactant in theMueller-Rochow Direct Synthesis is mostly sup-
plied by a recycle flow from the Methanolysis units.The deficit on chloromethane must be purchasedfrom external suppliers. The scenarios show a widevariation of external chloromethane supply. Theincrease of a factor of 1.2 in scenario A is only dueto a higher throughput of the Mueller-RochowDirect Synthesis while the significant increases inscenarios B and C, 3.2 times and 2.9 times, respecti-vely, are due to the change of the selectivity. Hig-her amounts of methyltrichlorosilane and methyl-hydrogendichlorosilane are produced in the Muel-ler-Rochow Direct Synthesis in these scenarios andin their further processing to precursors of resinsand organo-functional silanes no chloromethaneis created as a by-product feeding the recycle flow.Assuming higher costs for external chlorometha-ne supply compared to internally recycled this
P1: RawMaterials
1813 t
T1: Müller-RochowDirectSynthesis
P2: Waste
P3: Poly-mers, Oils
6944 t
P6:CH3Cl
2365 t
352 t
441 t
8317 t
555 t
T13: CH3ClRecycle
6389 t
5 t
T2:Distillation
T12: HClRecycle320 t
5546 t
2480 t
P5
4995 t
T4: Metha-nolysis
249 t
74 tT5: Methanolysis 2
1663 t
356 t
401 tT8: Alkoxylation
1414 t
T10: Hydro-silytion
3567 t
3908 t
P7: CH3Cl
207 t
1614 t
406 t
P10: HCL
1734 t
P8: Resins
P9: organo-functionalSilanes
116 t
effects directly the cost structure of all precursors.The variation of methanol supply in the diffe-
rent scenarios is mainly driven by the throughputsof Methanolysis 1 and Alkoxylation and its corres-ponding HCl-Recylce unit (T12). The demand on ace-tylene is directly related to the vinylmethyldichlo-rosilane throughput.
The benchmark should be continued with a costanalysis. Using extended material flow analysis thecalculation of conversion costs and contributionmargins deliver the necessary data base. Severalimportant factors can be included in this cost ana-lysis, e.g., fix and variable costs of all operation unitsand price elasticity of raw materials, utilities, ener-gies, and final products. Especially the allocationof costs at coupled productions must be conside-red.
After identifying valuable scenarios from a mate-rial and cost perspective the benchmark can beextended by technical feasibility studies and invest-ment estimations. This includes a bottleneck ana-lysis of all operations units and material transportsystems, for instance
Mueller-Rochow Direct Synthesis,- capacity of crushing and mixing devices- fluidized bed throughput defined by resi-
dence time and thermal household- cyclone, filter, and scrubber limits- capability of compressors, pumps and
evaporators, pipes and conveyors, etc.capacity and efficiency of distillation/sepa-ration,
- separation efficiency on different feed stocks (pressure levels, reflux ratio, fee-
Figure 4 Sankey diagram of scenario C with 100% throughput of the Mueller-Rochow Direct Synthesis with changed selectivity by using 50ppm lead and a converter to rearrange trimethylchlorosilane and methytrichlorosilaneto dimethyldichlorosilane.
ding points, etc.)- dimensions of columns and heat
exchangerscapacity of cycle flows,
- transport and interim storage- reconditioning and treatment
capacity of downstream units,- cycle times and reactor dimensions- capacity of post processing units, e.g.,
cleaning, packaging, etc.- operation planning of batch and semi-
continuous processes- interim storages
Configuration of utilities and energy suplies,e.g., heat integration via pinch technology,Environmental and safety matters, e.g.,limitson waste disposal or storage of hazardous mate-rials.
Based on the bottleneck analysis the necessa-ry investments for the debottlenecking can be esti-mated. Moreover a timeline and a transition planshould be developed how the current configurati-on of the integrated production network shouldbe transferred into the new configuration accor-ding to the investigated scenario. It is importantto consider the transition phase in the benchmarksince production should continue efficiently as longas possible during the transition which typicallytakes several years. Not all scenarios will have thepossibility to earn money during the transition intothe new configuration. The transition phase canbe modeled stepwise using several extended mate-rial flow analyses.
Important key figures can be estimated usingthe collected data of every scenario and benchmar-ked, e.g., ROI , IRR, earnings before interest, taxes,
Development of integrated production networks using extendedmaterial flow analysis
Figure 5 Mass flow of precursors in the various scenarios.
4000 t
3000 t
2000 t
1000 t
0 t
5000 t
Base Scenario Scenario A Scenario B Scenario C
Dimethylsilandiol
Trimethylsilanol
Methyldichloro-methoxysilane
Vinylmethyldi-chlorosilane
4212
t17
7 t
689
t34
8 t
4956
t20
8 t
811 t
410
t
3567
t20
7 t
1614
t17
34 t
3990
t0
t12
81 t
1734
t
depreciation and amortization (EBITDA ), etc.. A full benchmark can be done using a weighted
decision matrix where every scenario is scored alonga set of criteria or key figures (Grundig, 2006, Lunauet al., 2008). The weighting factor of every criteri-on or key figure is multiplied with the score andsummed up to a total score per scenario. The hig-hest total score is the best scenario according tothe criteria and weighting. The extended materi-al flow analysis is an iterative process. Any targetconflicts depicted in the benchmark may lead toideas or combinations of new promising scenarioswhich shall be investigated.
5 Discussion
In a simplified case study the methodology ofthe extended material flow analysis is demonstra-
ted. Even if the discussed drastic change in the pro-duct mix due to a change in the selectivity of theMueller-Rochow Direct Synthesis cannot be pro-duced without huge investments in different pro-duction lines, it demonstrates the strong effect offew process steps on the product portfolio and onthe economic efficiency of an integrated producti-on network. It also demonstrates the complexityand sometimes unexpected results of changes inthese networks. Especially the conversion cost ofthe final products in an adapted network estima-ted by cost analysis is of high value for far-reachingdecisions. In the discussed case further literaturesearch may reveal an adapted change of the selecti-vity of the Mueller-Rochow Direct Synthesis to bet-ter fit to the market demands and reduce invest-ments. Extended material flow analysis can be alsoused here to define the target for a R&D project to
Figure 6 Mass flow of raw materials in the various scenarios.
4000 t
3500 t
3000 t
2500 t
2000 t
1500 t
1000 t
500 t
0 tBase Scenario Scenario A Scenario B Scenario C
Raw Silicon
Chloro-methaneMethanol
Acetylene1541
t73
3 t
3293
t64
t
1813
t86
3 t
3874
t76
t
1813
t23
65 t
3263
t32
0 t
1813
t21
35 t
3336
t32
0 t
identify specific process conditions of the Mueller-Rochow Direct Synthesis to adjust the selectivityaccording to the defined optimized scenario.
Starting with material flow analysis and exten-ding with cost and investment analyses while usingscenario techniques, an optimized configurationof the integrated production network can be foundby using extended material flow analysis. The poten-tial to consider the majority of relevant aspectsdescribing the capability and efficiency of integra-ted production networks makes this methodologyvery powerful and beneficial. One modeled net-work can be analyzed from different perspectivessimultaneously, e.g., resource efficiency and pro-ductivity, economic and ecological efficiency, lifecycle assessment (LCA), technical capability, andrisk assessment.
The descriptive and prescriptive character ofthe extended material flow analysis allows esti-mation of unknown flows, identification of con-strains and correlations, cause-and-effect chains,dependancies between throughput, energy con-sumption, and costs providing a basis for productand project portfolio analyses, profitability analy-ses, and investment appraisals, to name only a viewpossible applications of extended material flowanalysis. Using various display formats, e.g., balan-ce sheets and Sankey diagrams, even complexresults of an investigation can be presented in atransparent and comprehensive manner. Therefo-re, extended material flow analysis supports thedefinition of proper mid to long-term developmentstrategies for integrated production networks tofollow the company’s vision.
References
Alberti, K.. (2001): Business Risk Management – StrategischerAnsatz zur Steigerung des Unternehmenswertes: in Fes-tel, G., Hassan, A., Leker, J., Bamelis, P. (ed.), Betriebswirt-schaftslehre für Chemiker – Eine praxisorientierte Ein-führung, Springer-Verlag, Berlin, p.109-115
Bode, A., Buerkle, J., Hoffner, B., Wisniewski, T. (2011): EinBlick durch die Kostenbrille: Anwendung von Stromana-lysen zur Bewertung und Verbesserung von Produkti-onsprozessen,Chemie Ingenieur Technik, 83 (10), p. 1553-1564
Brudermueller, M. (2001): Produktionsverbund in der chemi-schen Industrie: in Festel, G. Soellner, F., Bamelis, P. (ed.),Volkswirtschaftslehre für Chemiker, Springer-Verlag,Berlin, p. 168-180
Brudermueller, M., Langguth, E. (2001): Kuppelproduktion:in Festel, G., Hassan, A., Leker, J., Bamelis, P. (ed.), Betriebs-wirtschaftslehre für Chemiker – Eine praxisorientierteEinführung, Springer-Verlag, Berlin, p.288-298
Brunner, P. H., Rechberger, H. (2003): Practical Handbook of
Material Flow Analysis, CRC Press LLCDaenzer, W., Huber, F. (ed.), (1994): Systems engineering:
Methodik und Praxis, 8th ed., Verlag Industrielle Orga-nisation, Zuerich
Duggan, K. (2012): Creating Mixed Model Value Streams:Practical Lean Techniques for Building to Demand, 2nded., CRC Press, Boca Raton
Fandel, G., Fey, A., Heuft, B., Pitz, T. (2009): Kostenrechnung,3rd ed., Springer-Verlag, Berlin, p. 192-203
Gartner Inc. (2012): Cool Vendors in Green IT and Sustaina-bility, available athttp://www.gartner.com/technology/reprints.do?id=1-1AN8URJ&ct=120524&st=sg#h-d2e177 accessed Janua-ry 6th 2013
Grossmann, I. (2005): Enterprise-wide Optimization: A NewFrontier in Process Systems Engineering, AIChE Journal,51 (7), p. 1846-1857
Grundig, C-G. (2006): Fabrikplanung: Planungssystematik –Methoden – Anwendungen, 2nd ed., Carl HanserVerlag,Munich, p. 173-181
Haeuslein, A., Hedemann, J. (1995): Die Bilanzierungssoftwa-re Umberto und mögliche Einsatzgebiete, in Schmidt, M.,Schorb, A. (ed.), Stoffstromanalysen in Oekobilanzenund Oeko-Audits, Springer-Verlag, Berlin, p. 59-78
Langguth, E., Brudermueller, M. (2001): Kalkulation bei Kup-pelproduktion: in Festel, G., Hassan, A., Leker, J., Bame-lis, P. (ed.), Betriebswirtschaftslehre für Chemiker – Einepraxisorientierte Einführung, Springer-Verlag, Berlin,p.534-540
Lunau, S., Mollenhauer, J.-P., Staudter, C., Meran, R., Hama-lides, A., Roenpage, O., von Hugo, C. (2008): Design forSix Sigma + Lean Toolset – Innovationen erfolgreich rea-lisieren, Springer, Berlin
Moeller, A., Rolf, A. (1995): Methodische Ansätze zur Erstel-lung von Stoffstromanalysen unter besonderer Berück-sichtigung von Petri-Netzen: in Schmidt, M., Schorb, A.(ed.), Stoffstromanalysen in Oekobilanzen und Oeko-Audits, Springer-Verlag, Berlin, p. 33-58
Proud, J. (2007): Master Scheduling: A Practical Guide toCompetitive Manufacturing Scheduling, 3rd ed., JohnWiley & Sons, Hoboken
Schmidt, M. (2008a): The Sankey Diagram in Energy andMaterial Flow Management, Part I: History, Journal ofIndustrial Ecology, 12 (1), p. 82-94
Schmidt, M. (2008b): The Sankey Diagram in Energy andMaterial Flow Management, Part II: Methodology andCurrent Applications, Journal of Industrial Ecology, 12 (2),p. 173-185
Tomanek, A. (1992): Silicones and Industry. A compendiumfor practical use, instruction and reference, Carl HanserVerlag, Munich
VCI-Prognos-Study (2012): Die deutsche chemische Industrie2030, available at https://www.vci.de/Services/Publi-kationen/Broschueren-Faltblaetter/Seiten/PrognosStu-die_2030.aspx, accessed January 6th 2013
Viere, T. (2009): Verbundnetz chemische Industrie: in Schmidt,M., Lambrecht, H., Moeller, A. (ed.), Stoffstrombasierte
Development of integrated production networks using extendedmaterial flow analysis
