As for possible improvements, much information related to CHP and district heating in the former East block countries still rely on guesswork, but this should improve as traditions are created in those countries for collecting that type of data. Also, the estimation of the price elasticities of electricity in the countries must remain uncertain as in general only little experiences with price vs. demand behaviour of electricity have been obtained in the various countries up till now. Again, this should improve in the future. Finally, sensitivity analyses should be made of more parameters than it has been done till now to check how important the quality of the estimates are for different parameters, such as those related to elasticities.
Hence,
• Based on the problem description, a conceptual model of the modifications needed must be made and transformed into a mathematical model.
• The available version of the Balmorel model is modified with the extensions of the mathematical model
• The updated model is used for analysing the problem. It must be decided if further modifications are required to get a suitable representation of the model As expressed in Chapter 3, V&V of the modifications (i.e. conceptual model validation, computerised model verification, data validation, and operational validation) must be performed to ensure the model validity and also the model and model output appropriateness, as it should be assured that the questions are answered with a proper trade-off between quality, the computation time, and other resources needed, including the requirements of further data collection.
An example of a specific analysis that required modifications of the Balmorel model is presented in Paper E. This paper discusses the conceptual model development, the computer implementation, and presents the results of the analysis.
Not all modifications are easily made even with the flexibility of GAMS taken into account. The list below is ordered by how hard different types of modifications are (easiest first):
1. Changing existing data
2. Adding more information of the kind already available 3. Reducing the level of resolution
4. Adding new restrictions, new data, and/or new output routines 5. Changing of model type or changing the basic model structure
Ad. 1 – Changing existing data is the most easy modification. However, one must be sure that the new values entered are valid (i.e. efficiencies are between 0 and 1, consumptions are positive, etc.). The model itself does some checks for data validity.
Ad. 2 – Adding more information to tables already found in the model is almost as easy as updating the existing data assuming that the data to be entered has been found and are consistent with the existing (see previous section).
Ad. 3 – Reducing the resolution may many times be simple, but some problems may arise. For geography, exclusion of various countries changes the overall transmission network and excludes the import/export of the deselected countries with countries
outside the model. So while the reduction is easy to do, it may cause unwanted changes to the results. Reducing the resolution of time may be as simple, but often new profiles for demands, availability, etc. must be added. Finally, reducing the number of fuels or production technologies is difficult as they are interlinked (one cannot remove a fuel without removing the technologies using it) and the capacities of the existing power plants must be updated to represent the new system.
Ad. 4 – Adding new model restrictions, data, and output routines requires more work.
Even with relatively small and simple changes, validation of the modification itself and of its effects of the overall model must be made, while the previously mentioned changes in general should keep the model valid. An example of a modification belonging to this group is the already mentioned one presented in Paper E. Other examples are:
• Adding modelling of pumped storage power plants as presented in Paper B
• Adding more constraints on nuclear production, as it was found necessary for modelling the Lithuanian energy system; see Elkraft System et al. (2002) Ad. 5 – If one chooses to change the model type, for instance to introduce integer variables, or some basic model properties like the objective function, some major work is required. In general, the whole conceptual model must be validated again. Also, as the interpretation of the variables and the dual variables of the restrictions may be completely different and cause many of the output routines to be unusable. Still, a wide range of such “hard” modifications is possible within a modelling language as GAMS. Examples of such changes are:
• Changing the model from a deterministic linear program formulation to one using stochastic linear programming. It has already been discussed that the implications on the computation time is large, other solution procedures may be needed, and requires work on developing a scenario tree suitable for the problem.
• Adding integer variables, e.g. to include unit commitment decisions changes the interpretation of some results and again has large effects on the computation time.
• Changing the model, so it can use Cournot game theory for analyses of use of market power, the price of power must be part of the model formulation and not, as now, a result. Using a Mixed Complementary Problem formulation allows this, but requires many changes in the model formulation and the output procedures.
In conclusion, this section has shown that the Balmorel model has many possible uses.
However, it must be reckoned that models developed from scratch for a specific purpose may in relation to many of the model evaluation criteria, as those sketched in Section 3.4, perform better than the Balmorel model. Such models will on the other hand require time to develop, gather data, and get used to—time that may be considerably longer than the time needed for modifying the Balmorel model where one already may have the expertise running the model and interpreting the results. In addition, much of the data and the general relationships between the various elements of the power system that would be needed, are already present in the Balmorel model.