*5.8. Calculation of the Economical and Viable EFP of the Identified EFM*

As mentioned before, the final step of the proposed methodology, the calculation of the economic and viable EFP, based on the net revenues of the EFM was not a part of the performed analysis. Nonetheless, as can be inferred, the gross revenues are dependent on the variability of the price paid for electricity and will be specific to each MO. For MO-1 and MO-3, load increase revenues are achieved if the electricity price is lower than the average price paid for the combination of electricity and hot water as energy inputs. This consideration will further limit the active duration and the

activation frequency of these MOs, hence reducing flexible energy and increasing their specific costs. These considerations hint that these MOs might not be economically attractive for the company to activate. Nonetheless, they will be practically available if the EFM is implemented. On the other hand, for MO-2 and MO-4 load decrease, the revenues are reached if the price of generating HW is lower than that for electricity. This consideration will also reduce the active duration and the activation frequency of these MOs. Nonetheless, due to the considerable low specific cost of MO-2, and its high validity this might constitute a very attractive EFM overall for the company.

The economic EFP will hence constitute the flexible power and active duration in which the EFM generates revenues on each MO. The MOs that do not produce revenues should not be further considered. In the case of the viable EFP, the company has to make decisions on the active duration and activation frequency they intend for the EFM, weighing potential risks or negative consequences in the facility's performance, i.e., in its energy efficiency, which was not a part of this analysis.

## **6. Discussion**

The initial application of the methodology provided several insights that are discussed in this section. Under ideal conditions, the definition available industrial systems, Step 1, will only respond to the grouping energy consuming components in industrial systems and their categorization in technical units. Nonetheless, as detailed monitoring of all energy-consuming loads is not yet a standard in the industrial sector, a very relevant aspect in the decision of which industrial systems will be analyzed is the available data records of their energy consumption and their output. The selection of implementation objectives is relevant to provide an end goal to the analysis, however, it is frequent that once the EFMs are identified and characterized new implementation objectives become relevant.

The physical characteristics of the industrial systems, determined in Step 2, give a general view of the system and its operation and can lead to an initial understanding of the energy flexibility capabilities of the system. Nevertheless, excessive reliance on these characteristics might be misleading. During the application of the methodology, it was the case that initially thought available EFMs were deemed unavailable by the operative characteristics of the system or the production characteristics of the facility.

The suitability analysis, conducted in Step 3, allows sorting among the available industrial systems and reduce the duration of the analysis particularly when the production facility is very complex and hence constituted by a large number of industrial systems. Nonetheless, its qualitative nature demands caution as a wrong assessment of any of the three criteria might discard suitable industrial systems. In the cases where a determination was not clear, the consideration of operative characteristics of the industrial system, which normally occurs in the subsequent step, significantly helped the analysis.

In contrast to the physical characteristics, the operative characteristics of the industrial system and the production characteristics of the facility, determined in Steps 4 and 5, play a sorting role in either supporting or discarding the availability of each EFM-Category, as conducted in Step 6. Particularly in very wide encompassing EFMs categories, like dedicated energy storage, which initially seem to be available for all sorts of industrial systems.

The determination of the different parameters in the characterization framework in, Step 7, is intrinsically dependent of the nature of each industrial system and therefore is considerably difficult to standardize, here the experience of the person conducting the analysis, the thoroughness of the surveyed system and facility characteristics and, the input of relevant stakeholders from the production facility proved vital to obtain realistic values.

Similarly, the calculation of the economical and viable EFP, in Step 8, is very case-specific and only general guidelines can be given regarding how this step should be conducted.

In general, the application of the proposed methodology shows that it is not able to replace the accuracy of modelling the industrial system to simulate its operation under energy flexible operation, as described in References [15,38] among others. As explained, the methodology relies on typical profiles of energy consumption and patterns of operation. As these profiles and patterns are a simplification of

the actual dynamic operation of an industrial system, the performance of EFMs once implemented will diverge from the provided characterization. Nonetheless, the methodology presents considerable value, as it pinpoints the industrial systems suitable for energy flexible operation, from the large list of available industrial systems in a typical production facility. Moreover, it systematically identifies and characterizes the specific actions that induce energy flexible operation in these systems in the form EFMs, which is not only novel but provides a key input for the modelling, evaluation, implementation and management of the energy flexibility capabilities of the industrial system. Subsequent modelling of the industrial system acts then as a supplement, focusing on improving the accuracy of the values of the characterization parameters and being used as a prognosis tool to plan the management of the EFMs.

Additionally; the initial results also show that the methodology is promising but can be improved by improving the tools to establish the typical operative patterns of industrial systems. The accuracy of the results is highly dependent on the approach used to establish these patterns. It is hence crucial to examine thoroughly the available machine learning algorithms on data mining and clustering to find the best fitting for the task. These algorithms provide extremely relevant insights towards understanding how energy consumption is affected, particularly by the operative characteristics of the system and the production characteristics of the facility. Therefore, the most fitting algorithms and their optimal usage will facilitate the identification of EFMs and provide more accurate quantification of their characterization parameters.
