Water Safety Plan: A case study from Sofia, Bulgaria

According to Directive 2184/2020 on the quality of water intended for human consumption and World Health Organization (WHO) best practice, the Water Safety Plan (WSP) includes a review of the whole water supply chain – from protection of the water source, through water abstraction, transportation, treatment, storage and distribution to the end user. Risk management is based on the introduction of corrective, control and preventive measures, as well as subsequent validation of the measures taken and monitoring of their implementation according to the Deming cycle. The purpose of the current study is to review individual components of the WSP implementation and the effectiveness of WSP developed and implemented by Sofiyska voda JSC (SV) - the water supplier of Sofia, the capital of Bulgaria. The article presents the overall approach followed by the water supply operator, including the system review, risk identification and assessment, mitigation barriers, process control through critical control points (CCPs), review and verification of all risk management measures taken.


Introduction
The World Health Organization (WHO) has developed and recommended a tool [1] -water safety plan (WSP) manual to ensure availability of safe drinking water to the end user.The objectives of WSP are to prevent contamination at the source, throughout the treatment plant, storage and distribution system.As an addition, application of WSP helps/supports water supply companies and their efforts to manage, detect and be ready to react in case of an event that could deteriorate water quality and in the prioritization of the action plan and investments [2].These principles are also laid down in the recently updated legislation of the European Union -Directive 2184/2020 [3].
Since 2005, Sofiyska Voda JSC (SV) has been following its policy of continuous improvement by introducing international standards as in 2018, an integrated quality, safety and health at work and environment management system (IMS) has been effectively implemented in Sofiyska voda JSC according to the requirements of ISO 9001 [4], ISO 14001 [5], EMAS [6].The management of water quality and safety is part of integrated quality management system and is ensured by a number of procedures, instructions, registers, technological process diagrams and etc.Based on the established good practices regarding the adopted international ISO standards and those settled by the local IOP Publishing doi:10.1088/1755-1315/1305/1/012011 2 legislation -Regulation No. 9 on the Quality of water intended for drinking and domestic purposes (Regulation No. 9) [7] and European legislation (Directive 2020/2184) requirements, SV adopted a holistic approach in WSP development.
This study investigates the steps of WSP developed by the largest water supply company in Bulgaria -Sofiyska voda JSC, which supplies drinking water to over 1.2 million people.The purpose of this study is to review the main steps by the implementation of WSP and to assess the effectiveness, and the measures taken to control the risk and ensure the quality and safe drinking water to the end user.In Bulgaria no such assessment of water quality and safety risks has been performed prior to 2018.The novelty in this study is the holistic approach and complexity of risks evaluated.The article provides an example of the risk assessment and management of microbiological risks [8,9].It observes key microbiological parameters across the entire water supply system, starting from the main water source, Iskar Dam, through the Drinking Water Treatment Plant (DWTP) Bistritsa, and extending to the end consumer in the city center of Sofia.

The study area
All elements of the water supply system of the city of Sofia have been evaluated, starting from the catchment (Fig. 1a), passing through the water source (Fig. 1b), the drinking water treatment plants (DWTP), distribution network, storage reservoirs in the water supply network, end user (consumer) and mobile water tanks providing water when the "water supply" service is interrupted (Fig. 2).The company operates and maintains: a high mountain dam (Beli Iskar), Iskar Dam, river catchments in the Rila and Vitosha mountains and several captured springs in Vitosha, 5 DWTP, 59 reservoirs, 14 chlorination stations working with chlorine gas and 19 chlorination points and approximately 3,695 km of water supply network.

Methodology of risk assessment
Fig. 3 presents the methodology applied to develop the WSP.The methodology is based on the guidelines of WHO [1,8] and EN 15975-2:2013 Security of drinking water supply -Guidelines for risk and crisis management -Part 2: Risk management [12].The main stages have been met: assembly of a multidisciplinary team of experts with different experience, the system is described, hazards and dangerous events are identified, the risk is assessed, the measures to control and mitigate the consequences of the risk realization are described.The effectiveness of control activities has been validated.Each of the stages of WSP development is reviewed periodically and as circumstances change.For each stage of preparation there is a methodology that includes: Assemble water safety team (WST) -It consists of experts from various units of the company with knowledge and experience regarding the processes, equipment and dangers for water at every stage from water intake, purification, distribution in the water supply network to the end user -engineers, technologists, biologists, chemists, etc. WST reviewed regularly and updated documentation reflecting the circumstances changes, as necessary.The team includes both permanent members and experts for one-time consultations.Specific training on Hazard Analysis and Critical Control Points (HACCP) [13] was conducted to identify hazards and hazardous events.;Drinking water supply system description (DWSS) and Identification of hazards and hazardous event and risk assessment -WST described the system, identified hazards and hazardous events according to the WHO methodology [1,8,12], and the actual risk assessment was made according to an internal methodology used to assess all risks at company level.The sources of risks are divided into categories depending on what sphere they could arise in-changes in climate, geological, agricultural, industrial, wildlife, transport infrastructure, etc.According to the risk assessment [14], the risk assessment matrix is a 5-grade scale as follows: (i) impact (from low to serious); (ii) probability (from very unlikely to very likely) and (iii) effectiveness of risk mitigating measures (from noncontrolled to strong controlled).The lowest level of the risk is 1 and the highest level of risk is 125.
Risk control and management -In parallel with hazard identification and risk assessment, WST documents existing and potential control measures and assesses their effectiveness.Control measures are preventive and control, with the multi-barrier approach playing a leading role.The barriers are chosen in a way that, in case of possible removal of different stages of the water supply chain, duplication is secured.This approach provides sufficient reserve to continue operating under normal fluctuations in performance, which typically include periods of inefficiency.A multi-barrier approach [15] that exists "from the watershed to the end user" means that failure at one stage can be compensated for by the effective operation of the remaining barriers, minimizing the likelihood that pollutants will pass through the system and negatively impact customers.The risk reduction achieved by each control measure is an indication of its effectiveness.
An example of preventive measures are all related to the technical condition of the water supply facilities (river water intakes; water intake tower; supply water pipes; the building stock, technological security, etc.); ensuring service continuity -energy independence; energy and other utilities; the suitability of the equipment and its accessibility for cleaning, repair and preventive maintenance; management of materials in contact with drinking water (purification chemicals; filter materials, etc.); measures to prevent cross-contamination; cleaning and disinfection; pest control; personal hygiene of the staff.
Operational/process monitoring includes definition and validation of control measures and establishment of procedures to demonstrate that control activities continue to work.Critical control points (CCPs) have been defined for this purpose, with corresponding limits during the production and supply of drinking water to the end user through the network, which are subject to continuous monitoring to prove that they are under control.The system shall include planned measurements or observations related to the relevant critical limit.CCPs controlled in DWTP Bistrica are indicated in the operating instructions of the relevant facilities (Fig. 4).Control of processes along the entire water supply chain from the water source to the distribution network is carried out through SCADA (Supervisory Control and Data Acquisition).It is an automated dispatch control and monitoring system and consists of a central dispatch center, local control points, a guarding and monitoring system, a radio network for carrying telephony and telemetry data.Sensors are available in SCADA to measure quality parameters of raw, treated (drinking) and delivered drinking water.
Validation of the control approach and measures is the active acquisition of evidence that existing or new measures are suitable for control.The effectiveness of the control measures taken is confirmed by the independent control of the quality of drinking water at each stage of its production and delivery to the end user.The control is carried out by the accredited Laboratory Testing Complex at "Sofiiska Voda" JSC.Records are kept of the results of the inspections, which are issued in the form of monthly and annual reports on the quality of drinking water.The records are made available to all interested parties.The ultimate goal is compliance with the legal requirements for the quality of drinking water, defined in Regulation No.The eight monitored parameters were selected based on their versatility in interpreting the results as follows: Clostridium perfringens is Gram-positive, anaerobic, sulfite-reducing bacteria.They produce spores that are exceptionally resistant to unfavorable conditions, including chlorination.For its resistance and size (spores are smaller than protozoan (oo)cysts), Clostridium perfringens has been proposed as an indicator for effectiveness of the removal of protozoa by filtration processes [16].Total coliform bacteria include a wide range of aerobic and facultative anaerobic Gram-negative, non-sporeforming bacteria who can survive and grow in water, including supply system.Total coliform bacteria parameter could be used as a disinfection indicator.After disinfection, the quantity of bacteria has to be 0 and its presence indicates inadequate treatment [17]; E.coli and Faecal coliforms-are present in very high numbers in human and animal faeces and are rarely found in the absence of faecal pollution.E.coli and Faecal coliforms has been established as a most suitable indicator of faecal contamination and as a disinfection effectiveness indicator.Turbidity of the water is caused by finely dispersed organic and inorganic particles.Bacteria, viruses and protozoa are typically attached to particles, and removal of turbidity by filtration reduces microbial contamination in treated water.High levels of turbidity can protect microorganisms from the effects of disinfection [18].The most widely used units for measuring turbidity is Formazin Nephelometric Units (FNU) and can be seen by eye above approximately 4.0 FNU.To ensure the effectiveness of the drinking water treatment process, turbidity should be no more than 1 FNU and preferably much lower.Achievement of 0.3 FNU minimize the risk of pathogens that adsorb to particulate matter [18].The indicators COD and BOD5 were used as indicators for the total concentration of organics and for the concentration of biodegradable organics in the water source, respectively.Nitrate concentration is tracked in its role as a priority pollutant and in terms of risks associated with agriculture and blooms.

Categories and levels of risks in SV
The categories of risks identified are: i) risks for the water catchment and water sources, trunk mains; ii) risks in DWTP; iii) risks in the storage of water in reservoirs, iv) risks in the distribution network; v) risks by the end user/consumer; vi) risks in the transportation of water in tanks.In total, for all the 6 categories, 144 risks have been identified for the drinking water of the city of Sofia, with the most risks being in the DWTPs (55), watershed/water source/trunk mains (45), and the least being in the transportation of water in tanks (8) -Fig. 5.
As seen on Fig. 5 above, the most risks have been identified at the beginning of the water system, until the treatment process.After leaving the DWTP, the main risks for drinking water remain those related to unmitigated hazards that have not been eliminated at the previous level, as well as risks such as accident/disrupted water infrastructure; malicious acts; violation of the hydraulic integrity of the system; unmaintained indoor site infrastructure.For each of these risks, there are preventive and control measures.A typical example of a water network risk with a score of 10 (low) is an emergency, designated as "violation of the physical integrity of the network", which leads to a deterioration of water quality in terms of physico-chemical and microbiological indicators.The applicable control measures in the specific case are described in operational documents (procedures, instructions, plans, etc.), namely -Network management procedure; Agreement on levels of services provided by LTC to Drinking water supply network department; Service Agreement; Instruction for implementation of corrective actions in case of deviations in the quality of drinking water; Instructions for disinfection during repairs of sections of the water supply network; Procedure for "Good hygienic practices in the production and supply of drinking water".The risk assessment in SV varies between 1 and 125 and it is allocated into three categories (low 1-24; medium 25-49 and high 50-125).Fig. 6 presents the risks for Sofiyska voda JSC allocated by risk categories.It is noticeable that there are no high risks identified.The risk in the risk assessment matrix [14] is assessed after the control measures have been applied.This explains the fact that uncontrolled risks with a high rating are absent.Only 6 of all 144 identified and assessed risks have a "medium" rating, and all remaining 138 risks have a "low" rating.Some "medium" risks exists at the beginning of the process, in the components "water source" (2 out of 45 risks) and in DWTP's (4 out of 55 risks) -Fig.6.For example, few of the risks identified with "low" score at water source resulting in potential microbiological contamination, increased concentrations of pesticides and nitrates in the main water sources are: pollution, caused by agricultural practices (fertilization, plant protection, unregulated disposal of dead animals); faecal contamination caused by septic tanks; and illegal sewage from households and recreation places.The available preventive and control measures to mitigate the above-mentioned risks in the water source are: defined sanitary-protected zones I, II, III with the corresponding bans on some activities depending on the zone; hygiene (cleaning, sanitation and etc.); water intake from -25 meters; physicochemical treatment and disinfection in drinking water treatment plants (DWTP).The process of drinking water treatment and supply is controlled and managed using on-line monitoring with sensors (operational monitoring).The validation of the result of the undertaken measures is carried out on a daily basis by the accredited laboratory of Sofiyska voda JSC, which confirms the compliance of the drinking water by the consumer (verification monitoring).According to SV data, the end-user compliance for the period 2018-2022 is 99.79% on average for small and large areas.An example of a medium risk with a score of 25 in the DWTP is the difficulties with supply of chemicals -"The impossibility of obtaining supplies from manufacturers located outside the country, especially for critical raw materials and materials crucial to the purification process, becomes a significant concern when faced with the closure of state borders, such as the Covid-19 pandemic".If the risk is realized, the basic raw materials for the implementation of the process of purification and decontamination (disinfection) of drinking water on the territory of the Republic of Bulgaria will be missing.The control measures are: constant communication about "supply corridors" at the state administration level; active communication with the Ministry of Health; maintenance of a 30-day supply due to maintaining the disinfection of the DWTP outlet at an upper limit according to the "residual chlorine" parameter.

Verification of control measures
The risk management via WSP is based on a preventive strategy that encompasses the whole system to ensure safe drinking water.A key aspect of this approach is the implementation of water quality monitoring programs to confirm the effectiveness of the control measures, to verify compliance and to enhance understanding of system performance.An operational monitoring program to indicate whether individual hazard control measures (e.g. a treatment step) are working effectively and detects any deviation from required performance.For example, in 5 DWTPs, operational monitoring includes more than 40 sensors at the process (inlet/outlet) with local SCADA measuring following parameters: Temperature; Turbidity; Conductivity; pH; absorption units (organic); residual chlorine; organics concentration; number of particles (0.2-0.5 µm); pressure and twice per day for biggest DWTP taking of samples from accredited LAB for parameters, included in Regulation No. 9. A verification monitoring program is implemented to check the overall performance of the drinking water supply chain and the safety of drinking water being supplied to consumers.Verification monitoring involves periodical testing of water quality at the end of treatment, in the distribution network (reservoirs) and at the point of consumption by the end user using standardized analytical methods from accredited LAB.For example, 1715 samples with 38452 analyses from 86 points placed in Sofia city by the end user have been taken and analyzed for 2022.On top of that, about 1390 samples from 42 reservoirs in the distribution network were taken.On Fig. 7 the 86 sampling points by the end user are presented.
The monitoring points are representative of a given area and are evenly distributed throughout the city.Depending on the amount of water distributed in the respective settlement, the frequency of sampling can be from several samples per week to several samples per year.As proof of the effective operation of the multi-barrier approach described in the WSP along the entire chain from the watershed to the end user, several key (representative) water quality indicators -microbiological and physicochemical -have been selected, which are tracked and discussed below.
Fig. 8 presents the data for the investigated indicators for June 2020.It is observed that at the water source (CCP1 and CCP2) the quantity of the monitored microorganisms varies between 3 and 15 CFU/100 mL.In CCP1 in the surface sample, the amount of faecal coliforms was the lowest, followed by that of Clostridium perfringens and total coliforms.The presence of E.coli was not detected when examining the indicators.It is noticeable that microbiological indicators are higher in CCP2 -the sample taken from -25 m.The lower number of microorganisms in the surface sample is probably due to stronger sunlight and the bactericidal action of ultraviolet rays.According to Sigee, Clostridium perfringens is mainly present in the lower part (Hypolimnion) of the water column [27].C. perfringens does not multiply in water environment and probably, the detected singles cells could be spores or part of microhabitats with a lower oxygen concentration [17,18].In the following system components, it can be seen that 10 CFU/100 mL for Clostridium perfringens and 2 CFU/100 mL for Total coliforms were reported for the raw water at the inlet station.After the treatment of the water in DWTP Bistritsa at the outlet station (CCP 4) and in the water supply network (CCP 5), the research indicator groups of microorganisms, which show that the water meets the requirements for the quality of drinking water according to Regulation No. 9. Turbidity was measured in CCP3, CCP4 and CCP5, and it was found that it decreases the amount after the treatment of the water in DWTP Bistritsa and indirectly indicates the improvement of the water quality.
On Fig. 9 the data for the investigated indicators for June 2021 can be seen.It is found that the trend from 2020 continues with regard to Clostridium perfringens and E. coli.In 2021, the number of total coliforms is found to be slightly elevated (with 3 CFU/100 mL) compared to the previous year, which is negligibly low.Data on microbiological indicators and turbidity in the last two CCP along the water path (DWTP filtered water and supply network/end user) show that after the treatment of the raw water, the filtered water is already of potable quality and meets the regulatory requirements.Fig. 10 presents the data for the investigated indicators for May 2022.It is established that the number of total coliforms in CCP1 (the surface layer of water in the Iskar dam) in 2022 increased by about 6 times compared to the previous two years.This is probably related to the increased concentration of organics (measured as COD and BOD5) and nitrates, presented in Fig. 11a and Fig. 11b.The lack of significant deviations in the depth samples (Fig. 11b) in the number of microorganisms can be linked to the lack of similar large deviations in the values of COD, BOD5 and nitrates from the 3 years.
In 2022, the trend from 2021 for a higher number of total coliforms in the surface layer compared to -25 m is maintained (Fig. 10).The number of total coliforms is not considered a reliable indicator of fecal contamination according to some researchers [28,29,30].Panova [31] observed a positive correlation between the concentrations of chlorophyll a and pheophytin, and on the other hand, the numbers of total coliforms and the total count of heterotrophs, which (according to us) is due to the release of dissolved organic carbon that is readily available for uptake by microbial populations.As seen from the data in Fig. 11 in 2022, it was found that the concentration of organics (measured as BOD5) was increased 2 times compared to the sample from -25m.According to LeChevallier et al. [29] environmental factors (temperature and precipitation) and organic carbon concentration can be related to the occurrence of coliform bacteria, both in drinking water and in the water network.Water temperature is believed to be one of the main factors controlling bacterial growth, with coliform bacteria almost exclusively found in warmer water (above 15°C) in summer and early autumn.It was identified that during the sampling the water temperature in the surface layer was between 6 and 17°C, and in the -25 m layer the temperature was almost constant and varied around 6-8°C.According to the standards for quality requirements for surface water intended for drinking-domestic water supply [32], the amount of Clostridium perfringens is not standardized for surface water used for drinkingdomestic purposes.A limit value for Total coliforms of 50 to less than 5000 CFU/100 mL classifies the raw water as having high quality (category A1) and minimal need for treatment (rough mechanical treatment and disinfection, e.g.rapid filtration and disinfection).In the presence of more than 20 to less than 2000 CFU/100 mL faecal coliforms -the raw water is of high quality and is also classified in category A1.In each of the samples tested, the amount of faecal coliforms in the raw water was well below the contamination limit.For microorganisms, we can summarize that their amount in raw water is minimal, which guarantees raw water with excellent qualities, therefore less risk of contamination at the water source level.

Conclusions
The holistic approach in the implementation of the WSP is proof of the effectiveness of the measures adopted.The preventive and control activities undertaken throughout the drinking water supply chain, the process of continuous monitoring, both by online monitors and an accredited laboratory, the regular review of results, the monthly work of the water safety committee, communication and feedback to stakeholders, publication of the data ensure the optimal management and transparency of the monitoring process from the water source to the end user.The article examines a representative sample (single samples from the same May/June period in three consecutive years) of processindicative microbiological and physicochemical parameters that outline the general monitoring framework.The approximate number of analyzes by the accredited LTC per year is of the order of several thousand samples from the entire supply chain (water source, DWTP, distribution network, reservoirs, end user, water carriers.)The obtained data show effectiveness and efficiency of the applied multi-barrier approach in drinking water management waters in the city of Sofia.

Figure 1 .
Figure 1.Location of the watershed of the Iskar River on the map of Bulgaria (a) [10] and sanitary protected zones of the Iskar Dam (b) [11].

Fig. 2
Fig.2shows the scope of the Iskar water supply system (Iskar WSS), with the main water source being the Iskar dam (Fig.1), providing 80% of the water quantities needed for the water supply of the city of Sofia -the capital of Bulgaria with a population of 1.2 million inhabitants.3 of the 5 DWTP are located along the Iskar WSS.

Figure 2 .
Figure 2. Scope of water supply system Iskar, representative for 80% of water supply for Sofia city.

Figure 3 .
Figure 3. Water safety plan development approach with main aim, assurance of safe drinking water.

9 .
Critical control point (CCP) and monitored indicators -In order to track the effectiveness of the measures undertaken, 8 indicators (Clostridium perfringens, Coliforms count, E.coli, Fecal coliforms, Turbidity, Chemical oxygen demand -COD, Biochemical oxygen demand -BOD5 and concentration of nitrates) are reviewed in direct relation to biological and chemical hazards in 3 different years at 5 CCP, allocated along the entire water supply system from the water source to the end user.The selected points are: Iskar Dam -water intake tower, 0m (CCP1 -representative sample for the surface of the water source); Iskar Dam -water intake tower, -25m (CCP 2 -representative sample for the water intake horizon); DWTP Bistritsa, raw water (CCP 3 -inlet, representative sample for the raw IOP Publishing doi:10.1088/1755-1315/1305/1/0120116 water); DWTP Bistritsa, filtered water (CCP 4 outlet -representative sample for the treated drinking water); Supply network (CCP 5 -representative sample of drinking water quality at the end user).

Figure 5 .
Figure 5. Allocation of water quality and safety risks at 6 categories (Water source watershed/water source/trunk mains; DWTPs; Reservoirs; Supply network; End user; Water tanks).

Figure 6 .
Figure 6.Risks category divided by rating (low and medium) and part of the DWS.There are no risks with "high" rate.

Figure 7 .
Figure 7. Sampling points for drinking water by the end user in Sofia city.

Figure 8 .
Figure 8. Monitoring of indicators for the effectiveness of the applied multi-barrier approach in the supply of drinking water in the city of Sofia in 2020 in 5 CCP (CCP1, Iskar dam, 0m; CCP2, Iskar dam -25m; CCP3, DWTP inlet, raw water; CCP4 DWTP outlet, filters water; CCP5 drinking water network, end user).

Figure 9 .
Figure 9. Monitoring of indicators for the effectiveness of the applied multi-barrier approach in the supply of drinking water in the city of Sofia in 2021 in 5 CCP (CCP1, Iskar dam, 0m; CCP2, Iskar dam -25m; CCP3, DWTP inlet, raw water; CCP4 DWTP outlet, filters water; CCP5 drinking water network, end user).Fig.10presents the data for the investigated indicators for May 2022.It is established that the number of total coliforms in CCP1 (the surface layer of water in the Iskar dam) in 2022 increased by about 6 times compared to the previous two years.This is probably related to the increased concentration of organics (measured as COD and BOD5) and nitrates, presented in Fig.11aand Fig.11b.The lack of significant deviations in the depth samples (Fig.11b) in the number of microorganisms can be linked to the lack of similar large deviations in the values of COD, BOD5 and nitrates from the 3 years.In 2022, the trend from 2021 for a higher number of total coliforms in the surface layer compared to -25 m is maintained (Fig.10).The number of total coliforms is not considered a reliable indicator of fecal contamination according to some researchers[28,29,30].Panova[31] observed a positive correlation between the concentrations of chlorophyll a and pheophytin, and on the other hand, the numbers of total coliforms and the total count of heterotrophs, which (according to us) is due to the release of dissolved organic carbon that is readily available for uptake by microbial populations.

Figure 10 .
Figure 10.Monitoring of indicators for the effectiveness of the applied multi-barrier approach in the supply of drinking water in the city of Sofia in 2022 in 5 CCP (CCP1, Iskar dam, 0m; CCP2, Iskar dam -25m; CCP3, DWTP inlet, raw water; CCP4 DWTP outlet, filters water; CCP5 drinking water network, end user).

Figure 12 .
Figure 12.Water quality compliance levels for 2021 and 2022.

Fig. 12
Fig. 12 compared the levels of water quality compliance for 2021 and 2022.The values combine microbiological, physicochemical and radiological compliance data for drinking water quality, according to the current guidelines for calculating key indicators valid for Business Plan 2017-2021.The levels of compliance achieved during the period January 2021 -December 2022 exceed the business plan (BP) level for 2021/2022 of 99% for the key indicator compliance rate for large water supply areas, which is above the BP level for 2021 of 98.06% for the key indicator.During the period January 2020 -December 2022, no deviations were identified that represent a health risk for the population.