HBT Spotlight: Performance Meets Proof
“We rely on the multi-barrier approach” used to be a sufficient explanation in many discussions around drinking water safety. But now it’s the start of the conversation, not the end.
Across many utilities, the traditional response to microbial risk has been to point to the presence of multiple treatment barriers such as coagulation, filtration, disinfection and operational controls, and argue that the overall treatment train provides resilience. The concept remains sound. The issue is that regulators increasingly expect utilities to demonstrate how much protection each barrier contributes and whether that contribution can be supported with evidence. Under Chapter 5 of the Australian Drinking Water Guidelines (ADWG), the Health-Based Target (HBT) framework has shifted the discussion from “do we have multiple barriers?” to “can we justify and validate the microbial reduction each barrier claims to provide?”
Victoria has now moved to require water agencies to complete HBT assessments in accordance with ADWG Chapter 5, and similar expectations may emerge elsewhere soon. While implementation approaches vary between jurisdictions, the direction of travel is becoming clearer: microbial risk reduction is increasingly expected to be quantified and defensible. For many water agencies, this is changing the conversation around Log Removal Value (LRV) credits.
Moving beyond the traditional risk register
Many utilities already maintain risk assessments within their Drinking Water Quality Management Plans (DWQMPs) or Drinking Water Management Systems (DWMS). These should identify hazards, assign risk ratings and document preventive measures.
An HBT assessment supports this but asks more specific questions. Rather than asking whether microbial hazards exist and whether controls are in place, an HBT assessment asks:
What is the source and load of pathogens in the source water?
What pathogen reduction is required for the water to be considered safe?
What reduction does the treatment process currently achieve?
Can each claimed reduction be demonstrated and validated?
Where are the gaps?
That distinction is important because risk registers can sometimes become qualitative documents. They may identify filtration as a control measure and classify risk as “low” or “moderate,” but they may not quantify whether filtration is contributing 2 log, 3 log or 4 log reduction for a pathogen group.
Under Chapter 5, the emphasis moves toward quantifiable performance. For example, a source water assessment may indicate elevated pathogen challenge following high rainfall events or catchment disturbances. Under an HBT framework, this does not simply become an increased risk rating. It becomes an input into determining the microbial reduction required from treatment processes.
The discussion then moves to evidence. Can the treatment plant demonstrate that the required reduction is achieved under all expected operating conditions?
Building LRV credits under the HBT framework
LRV credits are built progressively across treatment barriers rather than being assumed collectively. A simplified treatment train may include filtration and chlorination. Each process may contribute a portion of the total microbial reduction requirement. Historically, some utilities relied heavily on indicative values from guidance documents or literature sources. In some cases, nominal filtration performance became accepted as equivalent to a particular credit level, even where validation evidence was limited.
Regulators are now increasingly challenging this approach. Common examples of questions include:
What validation evidence supports the LRV claim?
Were operational conditions representative?
Were challenge conditions adequately captured?
How were process failures treated?
Were periods of poor instrument performance excluded?
The issue is not necessarily that literature values are incorrect. Indicative values remain useful as starting points. The challenge is that they often do not reflect actual site-specific performance. Two plants using nominally identical processes can achieve different outcomes due to (but not limited to):
Source water quality
Seasonal variation
Chemical dosing strategies
Operational practices, including process interlocks
Asset and equipment condition
Instrument reliability
Consequently, site-specific evidence is increasingly becoming the expectation rather than the exception.
Where many LRV claims start to break down
In practice, the issue is often not treatment capability itself. The issue is evidence, or at times the right kind of evidence (e.g. continuous monitoring data). Many treatment plants operate effectively for years and consistently produce compliant drinking water. However, historical data sets frequently contain problems that become significant once they are examined through an HBT lens. Examples commonly encountered include:
Data gaps: periods of missing data for a process (e.g. filtration turbidity) may appear minor from an operational perspective but become difficult to justify during validation exercises.
Instrument flags: online instruments may generate alarms that may have been historically ignored operationally but become problematic when assessing long-term performance.
Undocumented calibration changes: changes in instrument settings, calibration practices or equipment replacement can affect confidence in historical trends.
Operational overrides: manual interventions and temporary changes to operating conditions may not have been formally recorded.
These issues can create significant challenges when attempting to use historical SCADA information to support LRV claims. A plant may have several years of archived data available, but if there are large periods where instrument reliability cannot be demonstrated, the usable validation dataset may become much smaller. The result is often unexpected. The treatment process itself may be performing adequately, but the evidence required to support the claimed performance is incomplete.
The SCADA data-cleansing challenge
Data cleansing is becoming a substantial component of HBT work. Historically, many utilities viewed SCADA systems primarily as operational tools. Their purpose was process control, alarms and trend visibility. Increasingly, SCADA data is also expected to support regulatory evidence. That changes the level of scrutiny applied to information such as:
Missing values
Signal dropouts
Instrument maintenance periods
Periods when the process is supplying water
Outlier removal methods
Data averaging assumptions
Time synchronisation issues
Questions that once had little practical consequence become important:
Should all data be considered?
Should outlier readings be excluded?
How should periods of maintenance be treated?
Was the instrument calibrated during the assessment period?
Was a sensor relocated?
Small uncertainties can accumulate and ultimately influence whether a claimed LRV is defensible.
Challenges for rural and remote systems
The transition toward HBT assessments can create additional challenges for rural and remote water supplies. Smaller systems may operate with limited staffing, fewer online monitoring instruments and less historical operational data than larger metropolitan utilities. In some locations, SCADA functionality may be limited or unavailable altogether, making it difficult to demonstrate treatment performance over the assessment period being considered.
In these situations, the challenge is often not that treatment processes are inadequate, but that evidence supporting treatment performance may be incomplete. Additional monitoring, improved record keeping or staged infrastructure upgrades may therefore become important considerations. The HBT framework may ultimately help identify these gaps, but it can also highlight where implementation pathways for smaller systems need to be targeted.
Practical questions utilities should ask
As regulatory expectations continue to evolve, a useful exercise is to review existing LRV claims through a Chapter 5 perspective. Questions may include:
Can every claimed LRV be linked to evidence? Evidence may include validation reports, operational data, manufacturer information and performance monitoring records.
Are assumptions clearly documented? Examples include assumed pathogen surrogates, filter performance assumptions (e.g. backwash, maintenance) and treatment process dependencies (e.g. location of instruments, flow meters, SCADA automations, interlocks).
Is historical SCADA information usable? Data quality problems frequently emerge during review. The earlier these issues are identified, the easier they are to address.
Are gaps operational or capital related? Some issues can be resolved through procedural improvements; others may require larger conversations.
Procedural improvements might include:
Better calibration records
Improved alarm management
Data quality checks
Documentation updates and record keeping improvements
Larger conversations might involve:
Installation of monitoring points, online analysers and SCADA
Instrument replacement
Process modifications
Additional treatment barriers (assets, e.g. a new filtration or UV step)
Distinguishing between these categories is important because not every gap automatically becomes a capital project.
The broader implication
The multi-barrier approach remains central to drinking water safety and continues to be a fundamental principle of the ADWG. What is changing is the level of evidence expected to support it.
Under Chapter 5, the discussion increasingly focuses on demonstrable performance rather than assumed performance. For some utilities, the largest gaps may not be treatment processes themselves, but the information systems and validation frameworks used to support them.
The practical question for treatment plant managers and drinking water quality leads is becoming increasingly clear: If each LRV credit in your DWMS or DWQMP was reviewed today under a Chapter 5 HBT assessment, could every claim be defended with evidence?
The answer may determine whether the next discussion centres on process optimisation, data quality improvements, or longer-term asset planning decisions.
Comments