MCDM - Questioning the SIMUS method

Ø What is SIMUS?

SIMUS (Sequential Iterative Modelling for Urban Systems), is a MCDM method launched in 2011.

It is owned by the School of Economics, National University of Córdoba, Argentina

Ø What does iterative mean?

It means that the mathematical algorithm, called ‘Simplex’, walks the contour of a polygon that contains all feasible solutions of the problem; this geometrical figure is formed by the intersections of the criteria. The optimal solution (Linear Programming Theorem), is in one of the vertices of the polygon

Ø Why SIMUS is said to be different to other MCDM methods?

§ SIMUS is structurally different regarding all other methods due to several reasons:

It is grounded on the geometrical representations of Linear Programming (LP) inequations, i.e., a simple problem can be solved either analytically or geometrically, and both reaching the same result. This can be seen in any book on Linear Programming. That is, you can see an optimal solution, but only in very simple scenarios, up to three alternatives. Beyond that it is impossible to represent graphically a problem with more than three alternatives, because we live in a tri-dimensional space, but analytically there is no limit to the number of alternatives and criteria

§ It is different because it does not need criteria weights, since the relative importance of each criterion is computed internally in each iteration. That is, in LP the relative importance of each criterion is not a constant, it is variable, and depends on the alternatives, consequently no subjective or objective alternatives are needed.

§ It is different because its basis: LP was created in 1939 by Leonid Kantorovich, who, for this development was awarded the Nobel Prize in Economics in 1956.

§ SIMUS is also different because, opposite to other methods, it follows the bottom-up procedure, that is, the DM can introduce his/her objections, modify values and include his preferences, only after a strictly mathematical solution is achieved. It means that the DM works on a very solid base

§ It is different in performing sensitivity analysis, since it works with increases/ decreases only on the criteria that are significant to the selected alternative. Instead of increasing/decreasing weights it works increasing/decreasing the amounts of the resources corresponding to each criterion, but restricted to the allowable range of variation of each criterion. Naturally, this step in mathematically supported; the only variable is the amounts the DM wants to change. Even if the DM exceeds the allowable range, the software automatically changes to another criterion

§ There is another difference: All MCDM methods can work with multiple objectives and with quantitative and qualitative criteria. LP and the Simplex algorithm only accept one objective and no qualitative criteria. These are two serious drawbacks.

This is the reason by which SIMUS was developed. Even when based on LP and the Simplex it can work with any umber of objectives and with any mix of quantitative and qualitative criteria. LP, as mentioned, produces optimal solutions while SIMUS delivers compromised or balanced solutions as all MCDM methods.

Ø Can results from SIMUS be validated?

No, they can’t, the same as in all MCDM methods. However, it has an interesting quality: SIMUS produces simultaneously two solutions, obtained by two different procedures, namely weighted criteria and outranking, and the results from both are identical. It does not mean that the result is validated; it only shows the strong mathematical basis of the method.

Ø What type of weighs does SIMUS use?

It does not use weights. The method computes the importance of each criterion automatically in each iteration, based on the ratios between the values of the selected alternative and the resources. That is, it is based in selecting the criterion that minimizes the use of resources

Ø Does SIMUS admit preferences from the DM?

Yes, in both the alternatives and the criteria. They can be in putted front (Top-down), or at the end (Bottom-up)

Ø Can it work with groups?

Yes

Ø In case of working with groups how does it harmonize the different opinions?

It works analyzing holistically the different opinions of each member of the group, whatever its number. The opinion of each one on data, i.e., the necessity of a new criterion, or even using preferences from each participant is uploaded in the initial matrix, all at the same time, the software run, and the final total result for the problem compared with the precedent.

If for instance, the analyzed criterion calls for maximizing, there is a global response, the Z value or project objective function, that must be greater than the precedent Z and vice versa, in case on minimizing.

In any case, if the contributions of the members of the group improves the result, in maximizing or minimizing, they are accepted, if not rejected. This procedure is repeated for all criteria, and consider that the analysis of criterion ‘n’ includes the result of its precedent criterion ‘n-1’. Therefore, the final result involves the opinions and preferences of the whole group and for each criterion, whatever their agreement ort disagreement.

Ø Can it work with clusters?

Yes. SIMUS can give the value of each cluster as a whole, as well as the value of each component of the cluster. In addition, it gives the value of relative participation of each cluster in the problem

Ø Is pair-wise comparison used in SIMUS?

No

Ø How are alternatives selected?

By analysing each alternative by itself and choosing the one with the best cost of opportunity

Ø Is SIMUS affected by Rank Reversal?

It can be in very few cases, precisely because it does not compare one alternative against another, only their respective cost of opportunity values

Ø What about normalization?

It comes with a drop-down menu with four different normalization methods for the DM to choose

Ø Are results different regarding the normalization used?

No, they are all identical, although there could be some variation when using max-min, because this methods tend to increase the distances between values, while for other normalization methods it is invariant

Ø In which language is SIMUS written?

In Visual Basic, but also is in the Java and Python libraries

Ø How much is the cost of the software?

SIMUS in its full power is free world-wide.

Just ask your zip copy to: [email protected]

Ø Which are the main characteristics of SIMUS?

Other than the technical characteristics already commented, its operative features are:

§ It applies the Simplex algorithm developed in 1948 by George Dantzig, nominated as one of the best algorithms of the 20th Century. The Simplex algorithm, under the name Solver, is in your computer as an Excel add-in since 1993

§ It can model precedence between alternatives, quantitative and qualitative criteria in any mix, maximum and minimum criteria in any mix, integers and decimals in any mix, resources and limits for each criterion, inclusive and exclusive alternatives, it can work with positive and negative values, as well as binary, in any mix. That is, in the same matrix integers, decimals and binary values are integrated

§ If the number of alternatives is too high it can reduce the number according the wishes of the DM

§ SIMUS delivers automatically:

- The solution, that is the value of each alternative

- The ranking of alternatives

- The list of criteria that influence the selected alternative

- The range in which these criteria can holistically change without altering the best solution, i.e., considers simultaneously the effect of changes in each criterion that affect the solution.

- An indication of in what percentage each objective is satisfied

- It warns the DM if the problem is unfeasible

- It is fast, a 20 x 10 matrix for instance can be solved in a few minutes. This allows DM to perform as many tests as he/she wants in reduced time.

18 – In which fields is SIMUS applied?

It works the best in complex scenarios. SIMUS is a normative method and so working for personal or naïve problems is not the best option

Known published papers per areas - at least for this writer - on projects and countries

where SIMUS was used

- Agriculture: Irrigation - Spain

- Agriculture: Selection of crops (Country name no available)

- Airport: Airport expansion- The Netherlands

- Assets management: Bridges, water trunks, gas lines, etc.- Canada

- Assets management: Forecasting works to be done in bridges in n a certain period-Canada

- Country development: Designing a composite indicator for country development – Spain

- Economics: Macro economic analysis - Argentina

- Energy: Determining best path for high voltage transmission towers -South Africa

- Energy: Renewable energy and transition – European Union

- Energy: Selection of sources and transition to sources free of CO2 (Country name no available)

- Energy: Waste to energy – Indonesia

- Energy: Transition to renewable energy – Canada

- Energy: Wind Turbines -Eslabón de cierre: Método SIMUS, Multi Criteria Optimización - Cuba

- Environment: Environmental indicators in a country- Canada

Government policies: Strategic environmental assessment of

energy policies – Canada

- Government policies: Macro economic analysis considering population, production, consumption, etc. - Argentina

- Health: Improving hospital care – Sri Lanka

- Infrastructure: Provision of basic urban infrastructure for run-down neighborhoods in five cities - Ghana

- International help: Selecting donors - Palestine

- Metallurgy: Mechanical industry and alloys – Cuba

- Mining: The Dual Paradigm of Mining Waste: “From Ecotoxicological Sources to Potential Polymetallic Resources”—An Example from Iberian Pyrite Belt (Portugal) – Portugal

- Mining: MCDM Applied to the Evaluation of Transitional and Post-Mining Conditions-An Innovative Perspective Developed through the EIT ReviRIS Project - Portugal

- Mining: Post mining landscaping projects - Portugal

- Railways: Evaluation of urban transport technologies – Bulgaria

- Railways: Transport plans for intercity passenger trains – Bulgaria

- Railways: Evaluation of alternative transportation policies for containers – Bulgaria

- Railways: Multi-criteria evaluation of railway network performance in countries of the TENT -T Orient -East med corridor – Ten countries in Eastern Europe, from the Black Sea to the Baltic Sea

- Railways: An integrated multi-criteria approach for planning railway passenger transport in case of uncertainty- Bulgaria

- Railways: Fuzzy-SIMUS Multicriteria decision-making method. An application in railway passenger transport planning – Bulgaria

- Railways: Analysis of policies for railway operator using SWOT criteria and the SIMUS method. A case for the Bulgarian railway network - Bulgaria

- Regional planning: Integrating a large city and its conurbation - México

- Rivers: Selection of best spots to measure water quality in the Niger River – Nigeria

- Rivers: Determination of most important pollutants and sites for the Niger River – Nigeria

- Site selection: Best locations for urban health centers - Argentina

- Social: People removal from shanty houses and relocation - Algeria

- Thermomechanical: Identification and measuring thermodynamics coefficients in refrigeration condensers-Spain

- Transportation: Determining best alternative in modal transportation – Indonesia

- Transportation: Selecting road routes using GIS (Country name no available)

- Transportation: Selection of fuel used for 18 wheelers long distance trucks - Spain

- Urban and peri urban: Integration of urban and peri urban areas - UK

- Urban and rural areas: Determining sustainability in urban and rural areas- Iran

- Urban assets: Planning scheduling municipal works in a certain period, like sewerage, water, hospitals, forestry, social issues, etc.- Argentina

- Urban mobility: People accesses– Spain

- Waste domestic: Selecting sites for garbage incinerators – Italy

- Waste domestic: Selection of domestic waste recycling policies - Canada

- Waste management: Domestic collection- Canada

Any comment, and of course, criticism, will be very appreciated