**4. IWA models sensitivity analysis review**

To ascertain the sensitivity analysis across the activated sludge modeling field, an extensive literature review was carried out. The review was based on sensitivity analysis in the IWA models including refinements of these. Model modifications possessing increased complexity were discarded. The logic behind the previous statement is that some of these refinements are still not understood in deep to put into a model for simulating scenarios. Also, research outputs that were not cited, excluding those published during the last year (2020) were not included, except for one article that concerned a project continuation. Consequently, guaranteeing only the inclusion of IWA models as well as cited papers.

### **4.1 Selection procedure**

The literature search was conducted on the Web of Science (WoS) database-also known as Web of Knowledge (WoK). As this review focus is to evaluate the sensitivity analysis conducted up to date around the IWA models the search query presented in Box 1 was used. The timeline chosen for the search was 2008-present, being the last search was run until 30 December 2020. This period was selected given the discussion taken place in the WWTmod2008 workshop about the importance of dealing with uncertainty for certifying model accuracy [11]. A summary of the conference proceedings surrounding WWTP modeling together with a

structure to identify sources of uncertainty within the facilities can be found at Belia et al. [11].

*Sensitivity analysis AND (Activated sludge model OR Benchmark simulation model OR MBR model OR BSM) AND (Wastewater OR membrane)* Covers all articles dealing with sensibility analysis applied to WWTP modeling according to the IWA models, including variations for MBR modeling and microbial process upgrades.

### **Box 1.**

*Search query (2008–2020).*

A total of 133 research outputs were found in the WoK database matching our search query (see **Box 1**) either in the abstract, title, or keywords. Hence, for determining the studies included within this review, a screening protocol was established for determining the articles' eligibility. Which was conducted by the main author. A graphical summary of the screening process is presented in **Figure 1**. Notice that it provides a classification of the included articles, those excluded by reasons as well as the subject areas and the total of research outcomes for those excluded due to the screening process.

**Figure 2** presents the flow diagram of the screening process followed during this review. It is noticeable that from the total research outcome (133), 88 were discarded due to mismatching the eligibility criteria. According to the information published in **Figure 1**, a total of 17 research groups studied different models such as aeration models MBRs, sludge dewatering units, among others. Even some of these publications conduct a SA, the researchers assessed the model biokinetics via Monod equations rather than using the ASMs. The articles concerning greenhouse gases (GHGs) emissions (14) were discarded as the formation of GHGs is a complex process not clearly enough understood to be modeled [4]. Moreover, now the major concern in developing countries is improving WWTP effluent quality rather than carbon footprint reduction.

As for the anaerobic digestion field (7) together with the one for industrial WW (9) were excluded due to the following. First, the anaerobic digestion itself is not an activated sludge process due to the lack of dissolved oxygen. Even so, plant-wide models as BSM2 consider anaerobic digestion processes but the main purpose is to assess the whole plant performance and the interaction between unitary processes (e.g., primary clarifier, bioreactor, etc.). Nevertheless, sensitivity analysis for these models have been published and might be useful for the interested [38–40]. Second, industrial WW usually possess high COD loads that must be treated anaerobically, and usually recalcitrant pollutants (7) are present, which are also beyond the scope

**Figure 1.** *Graphical summary of the screening process.*

**Figure 2.** *Research outcomes screening process flow diagram.*

of this review. While algae-bacterial processes (5) were discarded as still an emerging technology [41]. Finally, the rest were out of the scope (17) deemed as casualties of matching the search query or dealing with sensors, some were unclear, among other features.

As shown in **Figure 2**, From the 45 full-text articles assessed for eligibility, 13 were excluded due to the following reasons. Three (3) did not comply with the desired quality. Mostly because there seem not to be a justification about the scenarios being studied, there was no clarity in the results or provide bare conclusions, and the application of the ASMs was confusing. Additionally, these were not cited. Mind that only the articles published in 2020, rather for one exception, could have no citations given its recent publication. Another three (3) publications use the BioWin General Model (EnviroSim Associates Ltd., Canada)-also called BioWin AS/AD model, an extended version of the IWA Baker & Dold model, coupled with an AD process. Given the complexity of the model, the articles were excluded. Yet, if it is of the readers' curiosity, please refer to Liwarska-Bizukojc and Biernacki [42]. While the rest studied Monod kinetics or Metcalf and Eddy guidelines (4), the AS model was not clear (2), or its purpose was to include specific pollutants for assessing its degradation within an AS process (1). Consequently, a total of 32 articles were included within this review.
