Modelling Combined Heat and Power Plants: Modelling CHP Plants on a System Level in the EMPS Power Market Model
MetadataShow full item record
Combined heat and power ( CHP ) plants enable simultaneous production of electricity and useful heat allowing for high total fuel efficiency. 70 % of all electricity produced in Denmark in 2012 was produced from plants classified as CHP plants. Because of the close power market connection between Norway and Denmark, a sufficient modelling of the Danish production portfolio is important to Statnett, the Norwegian TSO.CHP plants are very complex to model at a system level as they participate in both power and heat markets and exist with such technological diversity. The objective of this thesis was to uncover potential for improvements and to implement new modelling elements to the modelling of Danish CHP plants in the SINTEF developed EMPS power market model. The EMPS model does not explicitly model heat markets.Three areas were found to have potential for improvements:1. The average annual production profiles: the existing production profiles were too volatile, seemingly random and lacked documentation2. The aggregation of small CHP plants: The existing aggregation of small CHP not sufficiently diversified to account for technological diversity at a system level3. Temperature dependent capacity: the CHP production was not temperature dependent apart from a general seasonal variability.It was assumed that CHP units operation can be modelled by a linear feasible operating region describing the relation between instant heat and power production. CHP utilities must meet the heat load at all times. Based on assumptions about the heat load as a function of outdoor temperature and historical temperature data, new annual production profiles relating to average temperature were created. A new method of aggregating small CHP plants was developed based on decentral DH utility statistics and a new way of determining their marginal cost. In addition, a function developed by SINTEF that corrects the CHP production capacity according to the actual temperature was implemented. The new modelling elements were largely based on a CHP operation strategy developed for this thesis.The new elements were implemented in steps to see the effect of each step. The implementations formed three new EMPS model datasets, in addition to the one for the pre-existing modelling. Each element was shown to have been implemented correctly and addressed the issues as intended.When comparing thermal production per week [GWh] from observed data with modelling results for the period 2001-2008 it was shown to be a trend that modelled thermal production follows the observed thermal production in general for all datasets. This is largely due to general, seasonal variations in available back pressure capacity at a low MC.However, the degree to which the data fitted with this trend varied amongst the model datasets. The new modelling elements proved to be incremental improvements with regards to following the observed thermal production from week to week. The pre-existing modelling performed worst and the new dataset with all three new modelling elements, NewModTC, performed best, with regards to matching observed thermal electricity generation. Introduction of the forced aggregated small CHP production was the most effective new modelling element to increase the R2 to indicate a better fit with the overall trend that modelled thermal production followed the observed.A comparison between observed data and results from the existing modelling showed that thermal electricity production in general was much more temperature dependent and less price dependent in reality compared to the model. The new modelling elements showed incremental improvements to the overall modelling, as thermal production became more temperature dependent and less price dependent, i.e. approaching the trends of the observed data. However, the comparisons also showed that there remains some work to increase temperature dependency and decrease price dependency for the modelled thermal production further.The new model datasets resulted in more volatile prices in Denmark on average across all scenarios compared to the existing modelling. The increased temperature dependency was the main reason for this. Implementing new production profiles changed the available, low cost back pressure capacity, so that less was produced, compared to the existing modelling, during high load hours, increasing prices, and more was produced during low load hours, contributing to decreased prices. It is likely, but not shown here, that this was due to a new, flat distribution of CHP production capacity over the week s 168 hours. Overall, the new small CHP aggregation resulted in a moderate price reduction, as production was forced at zero MC, increasing production especially during the winter. The function for temperature dependent capacity correction showed to change the prices for certain hours significantly, but no overall increase or decrease for neither initially low nor high price hours. The prices changed mainly due to the function regulating available back pressure capacity down, increasing prices, or up, lowering prices for individual hours.