Open AccessProceeding Paper

Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems^†

Catalina Ortiz

Fatemeh Hatam

and

Michèle Prévost

Department of Civil, Geological and Mining Engineering, Polytechnique Montréal, Montréal, QC H3C 3A7, Canada

Author to whom correspondence should be addressed.

^†

Presented at the 3rd International Joint Conference on Water Distribution Systems Analysis & Computing and Control for the Water Industry (WDSA/CCWI 2024), Ferrara, Italy, 1–4 July 2024.

Eng. Proc. 2024, 69(1), 84; https://doi.org/10.3390/engproc2024069084

Published: 9 September 2024

(This article belongs to the Proceedings of The 3rd International Joint Conference on Water Distribution Systems Analysis & Computing and Control for the Water Industry (WDSA/CCWI 2024))

Download

Browse Figure

Versions Notes

Abstract

Predicting Legionella concentrations reaching users through building water systems requires a comprehensive water quality modeling approach. We integrate various frameworks and data to test the effect of nutrient availability, temperature, chlorine, and biofilm interactions in modeling Legionella. We show that neglecting biofilm detachment underestimates concentrations up to 5.5 logs, while including it increases estimates by 4.2 logs. This study identifies critical factors and uncertainty sources for characterizing the Legionella fate and transport phenomena within buildings.

Keywords:

Legionella; EPANET-MSX; mechanistic modeling; building water systems

1. Introduction

Legionella propagation through aerosolized water from fixtures such as showerheads or faucets are an exposure pathway to waterborne disease in building water systems (BWS) [1]. Risk assessment models have been instrumental in estimating average infection risks under various exposure scenarios. Nevertheless, predicting spatiotemporal variations is essential for effective risk evaluation and development of mitigation strategies. To date, a sole study has used EPANET-MSX to simulate Legionella within a BWS [2]. Hydraulic-water quality models were fitted to 1-year data from a monitored water-efficient residence. The calibrated models show higher concentrations as consequences of increasing water stagnation, aligning to the expected trend of Legionella to proliferate with higher water age [1]. These results underscore the potential of EPANET-MSX to forecast exposure to Legionella. However, calibrated kinetic constants are only applicable to the particular system conditions. Considering nutrient sources, temperature, disinfectant, and biofilm interactions is suggested to improve the predictive capacity for Legionella [2]. Thus, in this study we test mechanistic modeling frameworks that incorporate these factors while pinpointing uncertainty sources and knowledge gaps.

2. Materials and Methods

2.1. Case Study

The case study residence hot and cold BWSs were built independently in EPANET 2.2 with water usage patterns in the kitchen sink and 2nd floor bathroom sink and shower [2]. EPANET-MSX files were created for every season using data from 58 discrete samplings as the boundary conditions in the heater and service line [2]. The .inp and .msx files were shared by the authors, jointly with field water quality data including temperature, dissolved oxygen (DO), free chlorine (FCL), heterotrophic plate counts (HPC), Legionella spp. (Lspp) measured by qPCR in gene copies (GC)/L, and dissolved organic carbon (DOC). The focus in this investigation is summer since the highest Lspp detection occurred during this season, totaling 12 data points from 25 June to 31 July 2018.

2.2. Modeling Frameworks

Microbial proliferation is described by four periods: initial lag, exponential growth, stationary phase, and death. The calibrated models employed a first-order growth equation constrained by the maximum observed concentrations [2]. Here, the net growth reaction of Lspp in bulk water (

C_{L E G, b}

) follows Equation (1), where

μ

and

k_{d}

are the growth and mortality rates. Four mechanistic models are tested by sequentially incorporating the effect of nutrients, temperature, chlorine, and interaction with biofilm (Table 1).

\frac{{d C}_{L E G, b}}{d t} = (μ - k_{d}) \cdot C_{L E G, b}

(1)

Nutrient limitation: The Monod equation is used to limit the maximum Lspp growth according to the substrate concentration, which is considered as DO and a fraction of DOC because the biodegradable part (BDOC) was not measured. Most parameters are taken for Legionella pneumophila (Lp) because no specific parameters fit to all species. Indeed, modeling only Lp would be more pertinent given its association with infections [1], yet no Lp was found in the case study network.

K_{O}

and

K_{s}

are approximated to the substrate uptake by heterotrophs [4,5] because no Lspp data were found. BDOC is modeled using the formulation and parameters from [4], where it reacts with FCL, facilitates bacterial growth, and increases due to cell lysis. Thus, the FCL and HPC calibrated models are utilized [2].

Temperature-dependent growth and inactivation: Temperature (

T

) is simulated as in [8] using measurements at the service line for the cold system, while a constant 60 °C at the heater in the hot system is used, as required by Legionella regulations, since no temperature data were available from the heater outlet. Temperature-dependent

μ

is computed using a Gaussian curve according to Lp growth rates at 18–47 °C [3]. For temperatures over 47 °C, first-order discrete death rates (

k_{d, t e m p}

) are used for the sum of culturable and viable-and-not-culturable Lp cells at 50, 56, 60, 65, and 70 °C [9].

Chlorination: The effect of chlorine is incorporated in

k_{d}

as the ratio between FCL and a log reduction

C t

value for planktonic Lp [6], thus dynamically determining the disinfection rate according to the FCL levels throughout the system.

Biofilm dynamics: Attachment to biofilm is considered negligible since Lp growth and detachment predominates in biofilms that have reached steady state [5]. Thus, attached Lspp concentration

C_{L E G, w}

is assumed constant throughout the network. The Lp detachment rate (

k_{d e t}

) fits a two-phase first-order decay because higher release occurs during the first 5 min and is lower for the following water use time [7]. The lower rate is applied to all flowing pipes and the higher rate only to pipes right before fixtures (distal pipes) in EPANET-MSX due to the inability to differentiate them according to water use time.

3. Results and Discussion

Compared to growth without restricting

μ_{m a x}

(“Maximum growth” as referenced in Figure 1), the Monod formulation effectively limits Lspp proliferation as a function of BDOC and DO. The nutrient-limited model represents the highest concentrations among the growth models implemented, being on average 1.2 logs higher in hot-water fixtures but 1.4 logs lower in cold-water fixtures compared to measured data. This might be due to restricted growth by low carbon levels entering from water mains, although the real DOC measurements were higher due to leaching from PEX pipes [2], which was not considered in the model. The hot system simulations align with expectations since Lspp inhibition occurs when water temperature exceeds 50 °C during water use, while stagnated water in distal pipes reaches temperatures conductive to growth (20–45 °C) [1] within 60–70 min. Interestingly, disinfection by low chlorine residuals in hot fixtures (<0.01 mg/L) control excessive growth during prolonged stagnation/low water usage (Figure 1a). Temperature in the cold system has a negligible impact because it is stable and consistently within the growth range, leading to identical results for nutrient-limited and temperature models (Figure 1b). Average chlorine of 0.24 mg/L entering from the water mains results in high residuals at fixtures and substantial Lspp inactivation across the cold network.

Before including biofilm detachment, predictions are up to 1.1 logs lower in hot fixtures (Figure 1a) and 5.5 logs in cold fixtures (Figure 1b) than measured Lspp. Notably, the models were calibrated assigning observed data to cold water in the shower because stagnated water reached ambient temperature quickly [2]. Modeling hot and cold water mixing would be more accurate, especially for showerheads where more aerosols are produced and there is a higher infection risk. The model including biofilm increases predictions by 4.2 logs in the cold system as a consequence of a large surface area from where biofilm detaches during water usage, approximating better elevated concentrations in first draw samples which have been attributed to detachment [10]. Yet, the model still fails to accurately predict observed Lspp levels, implying that improved estimations require more detailed modeling frameworks and finer parameter calibration.

4. Conclusions

Employing a mechanistic framework offers a practical method to expand the application of modeling endeavors across various systems. However, mathematically describing all chemical, physical, and microbial processes influencing Legionella persistence within BWSs is inherently challenging. Through the comparison of mechanistic models incorporating the effect of incoming nutrients, temperature, chlorine, and biofilm detachment, we demonstrate that reasonable approximations can be achieved. Yet, incorporating additional factors is still necessary for improved simulations, such as carbon leaching from plumbing pipes, hot–cold water mixing, chlorine diffusion into biofilms, and interaction with protozoan hosts providing nutrients and protection to intracellular Legionella. Ultimately, as the primary goal is to quantify the overall risk of exposure, accurately predicting elevated Legionella concentrations as a consequence of detachment necessitates determining and calibrating biofilm model parameters, especially attached concentration and detachment rates in distal devices that offer larger surface-to-volume ratios.

Author Contributions

Conceptualization, C.O., F.H. and M.P.; methodology, C.O., F.H. and M.P.; software, C.O. and F.H.; validation, C.O., F.H. and M.P.; formal analysis, C.O., F.H. and M.P.; investigation, C.O., F.H. and M.P.; resources, F.H. and M.P.; data curation, C.O., F.H. and M.P.; writing—original draft preparation, C.O.; writing—review and editing, F.H. and M.P.; visualization, C.O., F.H. and M.P.; supervision, F.H. and M.P.; project administration, M.P.; funding acquisition, M.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Industrial Chair on Drinking Water at Polytechnique Montréal—NSERC-ALLRP 560764-20.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

The original contributions presented in the study are included in the article; further inquiries can be directed to the corresponding author.

Acknowledgments

We extend our gratitude to Juneseok Lee (Manhattan College, Riverdale, NY) for sharing the original data sets, reviewing, and providing feedback on the results of this study.

Conflicts of Interest

The authors declare no conflicts of interest.

References

National Academies of Sciences, Engineering, and Medicine (NASEM). Management of Legionella in Water Systems; Nicholas, R.J.B.A., Ruth, B., Haas, C.N., LeChevallier, M.W., Letson, J.T., Prévost, M., Pruden, A., Swanson, M.S., van der Wielen, P.W.J.J., et al., Eds.; The National Academies Press: Washington, DC, USA, 2019; p. 304. [Google Scholar]
Palmegiani, M.A.; Whelton, A.J.; Mitchell, J.; Nejadhashemi, P.; Lee, J. New developments in premise plumbing: Integrative hydraulic and water quality modeling. AWWA Water Sci. 2022, 4, e1280. [Google Scholar] [CrossRef]
Hochstrasser, R.; Hilbi, H. The Legionella Lqs-LvbR regulatory network controls temperature-dependent growth onset and bacterial cell density. Appl. Environ. Microbiol. 2022, 88, e02370-21. [Google Scholar] [CrossRef] [PubMed]
Abhijith, G.R.; Ostfeld, A. Contaminant fate and transport modeling in distribution systems: EPANET-C. Water 2022, 14, 1665. [Google Scholar] [CrossRef]
Vincent, J.; Tenore, A.; Mattei, M.R.; Frunzo, L. Modelling drinking water biofilms: Bacterial adhesion and Legionella pneumophila necrotrophic growth. Commun. Nonlinear Sci. Numer. Simul. 2024, 128, 107639. [Google Scholar] [CrossRef]
Buse, H.Y.; J. Morris, B.; Struewing, I.T.; Szabo, J.G. Chlorine and monochloramine disinfection of Legionella pneumophila colonizing copper and polyvinyl chloride drinking water biofilms. Appl. Environ. Microbiol. 2019, 85, e02956-18. [Google Scholar] [CrossRef] [PubMed]
Heida, A.; Mraz, A.; Hamilton, M.T.; Weir, M.H.; Hamilton, K.A. Computational framework for evaluating risk trade-offs in costs associated with legionnaires’ disease risk, energy, and scalding risk for hospital hot water systems. Environ. Sci. Water Res. Technol. 2022, 8, 76–97. [Google Scholar] [CrossRef]
Zlatanovic, L.; Moerman, A.; van der Hoek, J.P.; Vreeburg, J.; Blokker, M. Development and validation of a drinking water temperature model in domestic drinking water supply systems. Urban Water J. 2017, 14, 1031–1037. [Google Scholar] [CrossRef]
Epalle, T.; Girardot, F.; Allegra, S.; Maurice-Blanc, C.; Garraud, O.; Riffard, S. Viable but not culturable forms of Legionella pneumophila generated after heat shock treatment are infectious for macrophage-like and alveolar epithelial cells after resuscitation on Acanthamoeba polyphaga. Microb. Ecol. 2015, 69, 215–224. [Google Scholar] [CrossRef] [PubMed]
Bédard, E.; Laferrière, C.; Déziel, E.; Prévost, M. Impact of stagnation and sampling volume on water microbial quality monitoring in large buildings. PLoS ONE 2018, 13, e0199429. [Google Scholar] [CrossRef] [PubMed]

Figure 1. Measured, calibrated [2], and simulated Legionella spp. through sequential inclusion of nutrient limitation, temperature effect, chlorine effect, and biofilm detachment at (a) bathroom sink (hot system) and (b) bathroom shower (cold system).

Table 1. Legionella spp. mechanistic modeling frameworks sequentially including the effect of nutrient limitation, temperature, chlorine, and biofilm detachment.

Model	$μ$	$k_{d}$	$+$
Nutrient- limited	$μ_{m a x} \cdot \frac{O}{K_{O} + D O} \cdot \frac{B D O C}{K_{s} + B D O C}$	-	-
Temperature	$μ_{\max} \cdot \frac{BDOC}{K_{s} + BDOC} \cdot \frac{OD}{K_{O} + OD} \cdot \exp (- \frac{{(T - T_{o p t})}^{2}}{{0.2 * (T_{m i n} - T_{o p t})}^{2}})$	$k_{d, t e m p}$	-
Chlorine		$k_{d, t e m p} + F C L / C t$	-
Biofilm		$k_{d, t e m p} + F C L / C t$	$k_{d e t} \cdot C_{L E G, w} \cdot A v$

μ_{m a x} =

3.85 × 10⁻⁵ s⁻¹: maximum growth rate from growth curves for a reference Lp strain grown in a minimum-nutrient medium [3].

K_{s} =

0.195 mg/L [4] and

K_{O} =

0.2 mg O₂/L [5]: BDOC and DO half saturation coefficients.

T_{o p t} = 37 ℃

T_{m i n} = 10 ℃

C t =

321 mg·s/L [6].

k_{d e t} =

1.0 × 10⁻³ s⁻¹ or 2.17 × 10⁻² s⁻¹ [7] only applied to flowing pipes,

C_{L E G, w} = 39 G C / {c m}^{2}

[7],

A v =

area per unit volume.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

© 2024 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

Share and Cite

MDPI and ACS Style

Ortiz, C.; Hatam, F.; Prévost, M. Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems. Eng. Proc. 2024, 69, 84. https://doi.org/10.3390/engproc2024069084

AMA Style

Ortiz C, Hatam F, Prévost M. Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems. Engineering Proceedings. 2024; 69(1):84. https://doi.org/10.3390/engproc2024069084

Chicago/Turabian Style

Ortiz, Catalina, Fatemeh Hatam, and Michèle Prévost. 2024. "Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems" Engineering Proceedings 69, no. 1: 84. https://doi.org/10.3390/engproc2024069084

APA Style

Ortiz, C., Hatam, F., & Prévost, M. (2024). Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems. Engineering Proceedings, 69(1), 84. https://doi.org/10.3390/engproc2024069084

Article Menu

Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems^†

Abstract

1. Introduction

2. Materials and Methods

2.1. Case Study

2.2. Modeling Frameworks

3. Results and Discussion

4. Conclusions

Author Contributions

Funding

Institutional Review Board Statement

Informed Consent Statement

Data Availability Statement

Acknowledgments

Conflicts of Interest

References

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI

Article Menu

Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems †

Abstract

1. Introduction

2. Materials and Methods

2.1. Case Study

2.2. Modeling Frameworks

3. Results and Discussion

4. Conclusions

Author Contributions

Funding

Institutional Review Board Statement

Informed Consent Statement

Data Availability Statement

Acknowledgments

Conflicts of Interest

References

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI

Uncertainty Sources in the Mechanistic Modeling of Legionella within Building Water Systems^†