Paper presented at "Systems engineering models for waste management"
International workshop in Göteborg, Sweden, 25 -26 February 1998.

The MARKAL systems engineering model for waste management

D.J. Gielen

ECN-Policy Studies
PO box 1
1755 ZG Petten
The Netherlands
Email: Gielen@ECN.NL

Readers can download the complete paper from the following site: http://www.ecn.nl/unit bs/etsap/markal/matter.html

Introduction

The MARKAL linear programming model was developed 20 years ago within the international IEA/ETSAP framework (International Energy Agency/Energy Technology Systems Analysis Programme). More than 50 institutes in 27 countries use nowadays MARKAL [1, 2]. MARKAL is an acronym for MARKet Allocation.

The model was originally developed for energy systems analysis. In recent years, the model has been extended for materials modeling. The model covers now the whole materials life cycle ‘from cradle to grave’.

This paper focuses on the pros and cons of regional systems engineering models like MARKAL for waste management. First the characteristics of the model are elaborated. The impact of waste disposal fees and greenhouse gas (GHG) emission reduction on European waste management are discussed in a case study.

General model description

A MARKAL model is a representation of (part of) the economy of a region. The economy is modeled as a system, represented by processes and physical and monetary flows between these processes. These processes represent all activities that are necessary to provide products and services. Many products and services can be generated through a number of alternative (sets of) processes. The model contains a database of several hundred processes, covering the whole life cycle for both energy and materials. The model calculates the least-cost system configuration. This system configuration is characterised by process activities and flows.

The database of processes and the constraints for individual processes and for the whole region are defined by the model user. Constraints are determined by the demand for products and services, the maximum introduction rate of new processes, the availability of resources, environmental policy goals for energy use and for emissions etcetera. Processes are characterised by their physical inputs and outputs of energy and material, by their costs, and by their environmental impacts. Emissions like CO₂, NO_x and SO₂ are considered. Other environmental impacts like waste volumes or land requirements can also be included. All environmental impacts are endogenised in the process costs and the costs of energy and matter flows between processes.

MARKAL is a dynamic model. The time span to be modeled is divided into nine periods of equal length, generally covering a period of 40 or 80 years. Within such a time horizon, technological change will be a major driving force for a changing systems configuration. Changing technology can be modeled through changing parameters in time for individual processes. Another option is the modeling of the future availability of new alternative processes. The model is used to calculate the least-cost system configuration for the whole time period, meeting exogenously defined product and service demands and meeting emission reduction targets. This optimization is based on a so-called ‘perfect foresight’ approach, where all time periods are simultaneously optimized. Future constraints are taken into account in current investment decisions.

MARKAL has orginally been used as an energy systems analysis tool. The modeling approach has been extended to materials system analysis ‘from cradle to grave’. Figure 1 shows the materials system model structure. All bulk material flows are included that are related to the end-use of materials and products in a country or a region. Full-scale energy and materials system MARKAL models are now available for the Netherlands [3] and for Western Europe [4]. These models have been developed for analysis of GHG emission reduction strategies.

Model objectives

"Systems engineering is the process of designing, or engineering, such systems with due regard to constraints and limitations in the degrees of freedom for the design, i.e., bringing the system into being the best possible way. Total systems engineering is implemented all the way, from a conceptual system early perception, to its final depletion" [5]. The MARKAL model is in this process a helpful tool in the system analysis and system design.

Regarding systems engineering for waste management, the model can be used for:

- analysis of future waste quantities and waste composition;
- analysis of the cost-effectiveness of waste handling processes;
- analysis of the impact of energy and GHG policies on waste handling.

The effects of materials substitution, changing consumer preferences, changing energy prices, and changing environmental policies on waste management can be analysed.

Comparison to other systems engineering models

A large number of systems engineering models exist, each with specific applications. The comparison in Table 1 focuses on environmental systems engineering models. i.c. LCA (environmental Life Cycle Analysis) and MFA (Material Flow Analysis), being the most widely used systems engineering tools for decision support in environmental policies. The system boundaries of different tools are elaborated in Figure 2. MARKAL can also be compared to Life Cycle Costing (LCC) systems engineering models, but this comparison is not elaborated in this paper. Compared to other optimisation models like MIMES [6] or DSS-DICTUM [7], the main differences are the consideration of materials storage in products and the detailed representation of electricity production and demand. It depends on the type of problem if this is a methodological advantage. The main advantage of MARKAL in comparison to other energy and material flow optimisation models is the availability of a worldwide set of databases and the expert support group ETSAP.

Table 1: Comparison of MARKAL, LCA, and MFA characteristics

MARKAL for whom and why ?

The MARKAL model is characterised by an exceptionally broad time horizon. As a consequence, it is generally used for long term research or investment decisions of national importance, e.g. for waste management, steel production, naphtha crackers or power plants with a life of several decades. The MARKAL complexity makes it most suited for assessment of decisions with significant impacts on the national level. The modeling approach is especially relevant in a case of a changing systems configuration. Due to the long life of capital equipment, such change can only occur over a period of decades. The MARKAL model is nowadays mainly used by national governments, national research institutes, utilities and large industries for decision support regarding RD&D planning and investments. Results prove to be useful on the aggregated national level and on the disaggregated technology/materials level.

Modeling software and modeling methodology

MARKAL is run at ECN on a conventional PC with a Pentium processor. The software requirements are GAMS, a matrix optimizer (e.g. OSL) and the MUSS system (MARKAL User Support System). The MARKAL model itself is freely available upon request. The other software must be acquired from the distributors. One model run for an average sized model requires 15-30 minutes of calculation time. The most costly step in the use of MARKAL is the laborious data acquisition, which may take several man-years for a full size national model. Fortunately, MARKAL models are nowadays available for many countries. Once such a database has been established, new processes and new scenarios can be analysed in a matter of hours or days.

MATTER: Western European MARKAL

Approximately one third of all GHG emissions can be attributed to the materials system. Changes in material flows show significant impact on GHG emissions and GHG emission reduction costs. The Western European MARKAL model has been developed within the MATTER project (MATerials Technologies for GHG Emission Reduction) in order to study these strategies in more detail. MATTER is a joint project of 5 Dutch institutes in the framework of the National Research Programme on Global Air Pollution and Climate Change (NOP-MLK). The final model version will be finished in the spring of 1998. The total development has required 20 man-years.

The model covers more than 25 energy carriers and 125 materials. More than 50 products represent the applications of these materials. 30 categories of waste materials are modeled. These materials are characterised by their physical characteristics and by their quality (e.g. steel scrap, demolition wood, polyethylene in municipal solid waste MSW).

An array of technological measures can be applied to reduce these emissions, ranging from fuel shifts in power generation and renewable energy sources to energy savings or shifts in materials use. For example GHG emission reduction strategies that have been considered in the materials system are:

Different reduction strategies influence each others efficiency. If for example the electricity production becomes less CO₂ intensive due to introduction of renewables, electricity production in waste incineration plants becomes less attractive for GHG emission reduction. As a consequence of such interactions, the assessment of the potential and the cost-effectiveness of reduction strategies requires an integrated approach that takes the whole energy system into account. MARKAL is especially suited to study such interactions.

Waste on the energy balance

Data for waste flows in Western Europe are not consistent. In [8] the amount of MSW is estimated to be 141 Mt in 1990. 34 Mt waste was incinerated in 1992 according to this source. 83% of the combustion capacity was equipped with energy recovery. The total MSW creation in Western Europe amounted to 225.3 Mt in 1993 according to [9]. 17% of this waste (38 Mt) was incinerated according to this source. The amount that is incinerated is similar according to both sources, but the amount of MSW differs. The difference is probably accounted for by a different definition of MSW. A recent analysis showed that different national definitions before 1994 are a major cause of inconsistent waste figures [10]. A proper comparison for 1994, based on consistent definitions, showed for 8 Western European countries MSW figures between 460 and 585 kg per person per year, with an average of 537 kg per person per year. Assuming this figure can also be applied to the other countries results in an estimate of 190 Mt MSW in Western Europe for 1994. This figure is in between both earlier estimates. Municipal construction and demolition waste not originating from households is excluded from the survey in [10]. Some of this waste may also be considered MSW in a broader definition. This narrower definition may explain the gap with the high estimate.

The energy content ranges from 9 to 13 GJ per tonne for individual countries. The MSW heating value is largely determined by the plastic content, the paper content, and the amount of kitchen waste. In some countries, separate collection and recycling for these flows has reached high levels. A typical MSW waste composition for Western Europe is shown in Figure 3. The problem with the use of this type of figures for modeling purposes is however again the unclear definition of MSW.

Another approach for estimation of waste quantities is based on a balance for individual materials:

Materials consumption - losses - stock increases in the product use phase = Waste arising

A combination of both approaches has been used for Table 2. For metals and for wood, the bulk of the waste does not end up in MSW. For the other materials, the bulk is accounted for in MSW statistics. Note that the definition of recycling in Table 2 differs from the definition in some other statistics in order to generate comparable data for different materials. The definition that is applied here is the amount of waste input into production processes divided by the waste arising. The waste arising is defined as the total of recycling, incineration, and disposal (this excludes all kinds of ‘losses’ and excludes the increasing materials stock in the use phase).

For some materials, the amount of waste is considerably lower than the materials consumption due to net exports of semi-finished and finished products, increasing product stocks, and losses in the use phase due to oxydation etc.. The heating value of the total waste flows is indicated. The total energy content is approximately 2700 PJ. Total Western European primary energy use is approximately 55,000 PJ per year. Comparison of both figures shows that waste materials can cover 5% of the primary energy consumption. However, recycling makes often more sense, as both feedstock energy and process energy use can be reduced. The selection of the best strategy from an environmental and cost point of view depends on the policy goals and the systems configuration. This will be elaborated for one group of materials. Table 2 shows the highest disposal fraction for plastics. The following analysis wil focus on the potential for GHG emission reduction through energy recovery and recycling of plastic waste.

Table 2: Waste balance for important groups of materials, Western Europe (EU+EFTA), 1993/1994 [12, 13, 14, 15]

Case study: the impact of GHG emission reduction policies on long term plastic waste management

In the so-called 'stair-concept' of waste management; prevention, recycling, incineration with energy recovery and disposal are considered progressively worse options [16]. EC (and national) policies strive for improvements in the present waste management situation, that is dominated by disposal (Table 2). Incineration and recycling are rapidly evolving. However for many plastic waste types today, recycling is not cost-effective with state-of-the-art mechanical recycling technologies, since no markets exist for the low-quality materials that are produced.

Options for plastic waste handling have been characterised in detail [17]. Each technology has specific requirements for waste quality and energy inputs. The energy efficiency and the costs of some energy recovery options for plastics that are currently discussed are shown in Table 3. A comparison of the energy efficiency figures suggests that cement kilns and blast furnaces are the most efficient applications. However, a proper analysis must compare the energy efficiency of competing processes that can deliver the same products.

Table 3: Comparison of future processes for energy recovery from plastic and rubber waste [17, 18, 19]

The question of cost-effectiveness of CO₂ reduction through plastic waste management requires a more detailed approach. Ethylene, polyethylene (PE), polypropylene (PP), polystyrene (PS), polyvinylchloride (PVC) and other plastics are modeled separately. Figure 4 shows the MARKAL modeling approach for polyethylene. Five feedstocks options for ethylene are included. Polyethylene is used as input for many product categories that are all separately modeled (e.g. packaging, cars, appliances, etc.). Plastics compete with other materials for these markets. Waste occurs during assembly, use and removal of products, and the characteristics of plastic waste vary accordingly with processing technology and material application. Three kinds of plastic waste are discerned per polymer type:

Each plastic waste management technology requires a certain minimum input quality (see Figure 4). Eight types of plastic waste management options are modeled:

Only grate firing has been modeled as electricity producing technology. Table 3 lists a number of alternatives (NOELL, Lurgi, Siemens technology). However, their net efficiency is lower, while costs are higher. A second consideration is that grate firing is currently the dominating energy recovery technology.

There is no problem in handling high-quality waste in technologies that require low-quality waste inputs (e.g incineration), but the other way around is impossible without an upgrading effort. This is modeled like a cascade, represented by ‘dummy’ processes that convert high quality waste into mixed plastic waste and that convert mixed plastic waste into plastics in MSW. Upgrading is modeled as a sorting process that converts plastics in MSW into 'mixed plastic waste'. The costs for this sorting process step in the German DSD (Duales System Deutschland): DEM 2.61 per kg mixed waste [20]. Future costs will probably decrease. 50% lower costs are assumed by the year 2000, and further cost reductions beyond 2000.

1. Re-extrusin

Re-extrusion technology is only applicable for high-quality waste types. The plastic is ground and extruded. The material output quality depends largely on the waste input quality, compatibilizers can improve the output quality [21, 22]. If mixed plastics are used as input, the resulting material is only suited for a limited number of applications.

2 Solvent separation

The process is based on the difference in solvability of plastics in organic solvents. The process uses selective dissolution at increasing temperatures and flash devolatilization to separate mixed plastics into component polymers with pigments and fillers predominantly removed. The process has been developed on a pilot plant scale, it is uncertain what results will be achieved on industrial plant scale [23, 24].

3 Pyrolysis DRP process (mixed plastic waste input)

Pyrolysis is the process where hydrocarbons are heated in an oxygen-free atmosphere. At a temperature of several hundred ⁰C, the hydrocarbons decompose to yield a mixture of ssolid, liquid, and gaseous products. The product composition depends on temperature and pressure. The higher the temperature, the more gaseous products. An important fraction of this gaseous product is ethylene, if plastics are used as feedstock. Pure ethylene is the most valuable product, that can be used for plastics production. The ethylene yield results vary considerably, references state a yield of up to 40%. Such a high value is not yet proven on a large scale. In this study, a lower ethylene yield value has been applied. 25% of the LHV of the plastic waste is estimated to be used for process heating purposes. Plastic pyrolysis technology is tested on pilot plant scale. Coke and oil by-products faced quality problems in the past, the present status is unclear [25].

4 Hydrogenation VEBA process (mixed plastic waste input)

Plastics can be treated with hydrogen to produce feedstock like a naphta-like product and a hydrogenation residue, that can be used in coking processes. In Germany, a pilot plant exists and the construction of a large-scale plant is on it's way. The technology can be characterised as a thermal hydrocracking/hydrogenation process. The reactions take place in a liquid phase reactor and a gas phase reactor at temperatures of 400-450 ⁰C and a pressure of up to 250 bar. The main problem is currently the feeding of plastics into the reactor. Data for hydrogen consumption are still uncertain, but seem to be significantly higher as might be expected on the basis of the plastics chemical structure [26].

5  Plastic waste injection in blast furnaces (mixed plastic waste input)

Plastic waste injection into blast furnaces is currently practiced on pilot plant scale in Germany [27, 28]. Cost data have not been encountered. Because the additional equipment is similar to the equipment for incineration in cement kilns, the same cost data have been applied.

6  Plastic waste incineration in cement kilns (mixed plastic waste input)

Incineration of waste types in cement kilns is widely spread over Europe. Plastic waste incineration in cement kilns has been developed in Italy. The application of plastic waste requires special waste injection equipment due to its light weight. Investment costs for storage, transportation, and injection equipment are approximately 40 ECU/t plastic waste capacity. Annual costs for labour etc. are additionally 10 ECU/t [29].

7  Incineration grate firing (MSW input)

Current grate firing systems achieve an efficiency of 20-22%. Higher efficiencies are possible if the incineration plant is coupled to combined cycle power plants. LT steam from the incineration plant is further heated in the power plant and subsequently used in a steam turbine. Such combined plants can achieve a 28% efficiency for the incineration section. One such plant has been built in the Netherlands and is currently operating [30].

8  Disposal (MSW input)

Disposal costs are largely determined by government intervention. Large differences exist between countries. In the calculations, it is assumed that the disposal costs increase from 50 ECU/t in 1990 to 150 ECU/t in 2010 and subsequently to 200 ECU/t in 2040.

Three scenarios are investigated. A base case (BC) without emission penalties and two GHG emission penalty scenarios. These scenarios are shown in Figure 5. They are characterised by their ultimate penalty level of 100 and 200 ECU/t CO₂ equivalent, respectively.

The heating value of MSW will significantly increase due to a changing waste composition (Figure 6). These changes are caused by changing consumption patterns and changing waste management practices. Recycling is still increasing for paper, waste wood is increasingly separated for recycling and for energy recovery. Kitchen waste is recovered and used for compost production through aerobic and anaerobic digestion. What remains is plastic. The heating value of waste plastic is however much higher (30-40 GJ/t) than the heating value of the natural organic fraction (10-15 GJ/t). If no measures are taken, this development will cause serious problems for grate incineration installations.

The development of plastic waste management in the base case is illustrated in Figure 7. Disposal is graduallly replaced by incineration. Recycling does also increase, but does not reach the high recycling rates like for metals or paper. The amount of plastic waste still increases due to increasing consumption.

Figure 8 shows the results for the base case and the emission penalty scenarios in 2020. The total amount of waste is hardly affected due to these penalties. However, the waste management technologies will be significantly affected. Disposal does slightly increase and reaches its upper bound. As a consequence, recycling technology is introduced on a large scale: incineration is replaced by hydrogenation.

Table 4 shows the price developments in the plastic chain due to changing GHG penalties. Especially the coal price is significantly affected, but other prices do also significantly change. It is surprising to see a price decrease for ethylene and polyethylene. This decrease is caused by a combination of the carbon accounting practice and technology selection. The calculations show a shift to biomass feedstocks for ethylene production. The feedstock carbon in plastic production is modeled as carbon storage, followed by carbon release in the waste stage, in line with the IPCC guidelines [31]. The benefit of carbon storage for biomass based plastic production results in a net price decrease. This result shows how future carbon accounting practices can affect the materials prices and the materials competitiveness. Such effects must be considered in future emission reduction policies.

Table 4: Prices of materials and waste materials, base case and 100 ECU/t CO₂ penalty case, 2020 (negative price = net waste handling costs)

Table 5 shows a comparison between incineration and hydrogenation costs in the base case and the 200 ECU/t CO₂ equivalent case for plastic waste treatment. The figures show that significant extra costs occur if the environmental impacts are endogenised. The value of the products changes for both incineration and hydrogenation (see also Table 4). The impact of CO₂ reduction on the economics of both waste management routes is significant. Both the changing product prices and the emissions of the incineration do significantly affect the net costs of both waste management technologies. As a consequence of the lower costs for hydrogenation in the caes with emission reduction penalties, incineration is replaced by hydrogenation.

Table 5: Comparison of costs for incineration and hydrogenation of plastic in MSW in 2020

In a sensitivity analysis, the disposal fees increase from 50 ECU/t in 1990 to 250 ECU/t in 2010 and to 375 ECU/t in 2040. Comparison of this sets of calculations to the first set of calculations showed that this 50% increase of disposal fees results in a 10% shift from waste disposal to more incineration.

Conclusions

The MARKAL model has been successfully extented from energy systems modeling to integrated energy and materials system modeling.

Compared to other systems engineering models for waste management, the advantages are:

• a straightforward comparison of waste management strategies on the basis of life cycle costing, including endogenised environmental impacts;
• the impact of changing GHG policies and energy policies on the future waste quantity and waste composition can be analysed;
• the impact of changing GHG policies and energy policies on the environmental perfomance of waste management strategies can be analysed;
• the impact of a changing reference energy and materials system on the environmental impact of waste management strategies can be analysed;
• the result is one optimal solution;
• the cost-effectiveness of environmental policies that affect waste management can be compared to other strategies to achieve the same results.

The disadvantages of the MARKAL approach are:

• the analysis is straightforward if existing models can be used. The method is very laborious if full size models must be developed;
• the model is especially relevant in situations with changing technology and a changing reference system. Due to the long life of capital equipment and the slowly changing lifestyles, such changes take decades. Within a time horizon of one decade, the reference situation will not significantly change, and a straightforward chain analysis method will generally suffice;
• being an LP model, non-linear relations cannot be modeled;
• the quantitative analysis of the rationality of the model calculation results, especially life cycle emissions, requires laborious additional calculations;
• the model is not (yet) suited to analyse environmental problems like toxic emissions.

The energy content of waste materials constitutes approximately 5% of the total primary energy consumption in Western Europe. Waste combustion poses an attractive strategy to recover this energy. However, this waste management option is a source of GHG emissions. As a consequence, future GHG emission reduction strategies will have a major impact on the selection of future waste management strategies. The MARKAL model is especially suited to study the cost-effectiveness and environmental impacts of different waste management strategies for different GHG emission reduction scenarios.

The case study for plastic waste management in Western Europe shows that the future waste management is significantly affected by GHG emission reduction scenarios. GHG emission reduction has only a limited impact on the waste composition. However waste prices are significantly affected. The attractiveness of grate firing decreases. There will be increasing pressure to shift to disposal. If this option is blocked by a ban on disposal or by increasing disposal fees, grate firing is substituted hydrogenation.

Footers:

1) Paper prepared for the workshop "Systems engineering models for waste management", 25-26 February 1998, Göteborg.

2) Includes anaerobic digestion (for food/garden waste) and recycling abroad (e.g. for textiles)

3) Both with and without energy recovery

4) Costs includes investments, maintenance, residue disposal. Costs/gate fees for waste and/or revenues for electricity and heat are excluded

5) Incinerator cost allocation for incineration of MSW on the basis of heating values

6) Based on the 5% discount rate

7) CO2 emission must be accounted for hydrogeneration because emissions for the reference production are accounted on the mining/import of fossil fuels instead f accounting during final use