Water Quality Troubleshooting & Trends

Learn how operators troubleshoot water-quality problems using trends, baselines, related parameters, rate of change, process lag, instrument verification, source conditions, treatment data, and distribution-system information.

Water-quality troubleshooting is the process of turning measurements, observations, and operating data into a reasonable explanation for what is happening in the system. The most important skill is not memorizing one cause for one symptom. It is learning how to compare related information and eliminate unlikely explanations.

A good operator asks several questions before changing the process: Is the measurement real? Where did the change begin? How quickly did it occur? Which related parameters changed with it? Is the source changing, is treatment losing control, or is the problem developing after treatment?

Start with a Baseline

A baseline describes normal water-quality behavior under expected operating conditions.

A useful baseline can include:

normal range;
seasonal range;
daily pattern;
typical response to storms;
typical response to source changes;
normal treatment settings.

Why Baselines Matter

A result cannot be called unusual unless the operator understands what is normally expected.

For example, a temperature of 50°F may be completely normal in winter and unusual in midsummer.

One Number Is Not a Trend

A single result can represent:

real process change;
temporary variation;
sampling error;
instrument error;
data-entry error.

Trend data provide context.

Trend Direction

Operators should identify whether a parameter is:

stable;
increasing;
decreasing;
cycling;
changing suddenly.

Rate of Change

The speed of change can provide important clues.

A rapid change may suggest:

source switch;
chemical-feed failure;
instrument problem;
main break;
spill or unusual discharge.

A gradual change may suggest:

seasonal conditions;
increasing water age;
chemical depletion;
slow fouling;
changing source blend.

Percent Change

A simple percentage-change relationship is:

Percent Change = (New Value - Old Value) ÷ Old Value × 100

Percent Change Example

Raw-water turbidity increases from 10 NTU to 15 NTU.

Percent Change = (15 - 10) ÷ 10 × 100

Percent Change = 50%

The increase is 50 percent.

Do Not Let Percent Change Hide the Actual Magnitude

A large percentage change in a very small number may still have limited operational significance.

Always review both:

absolute value;
percentage change.

Averages

An average can help summarize a group of results.

A basic arithmetic mean is:

Average = Sum of Values ÷ Number of Values

Average Example

Residual measurements are:

1.2 mg/L;
1.0 mg/L;
0.9 mg/L;
1.1 mg/L.

Average = (1.2 + 1.0 + 0.9 + 1.1) ÷ 4

Average = 1.05 mg/L

Averages Can Hide Important Extremes

An average may look acceptable while individual measurements include:

very low values;
very high values;
large variability.

Review minimum, maximum, and individual results as well.

Rolling Averages

A rolling average uses a moving group of recent measurements.

It can help smooth short-term noise and reveal longer trends.

Do Not Smooth Away a Real Event

Rolling averages can make rapid changes look smaller.

Operators should still review the original measurements.

Variability

A process can have the same average value but very different stability.

Increasing variability can indicate:

poor control;
intermittent equipment problems;
changing source conditions;
unstable hydraulics.

Compare Similar Time Periods

Useful comparisons may include:

today versus yesterday;
this week versus last week;
this month versus the same season last year;
high-demand versus low-demand periods.

Seasonal Patterns

Many water-quality parameters change seasonally.

Examples include:

temperature;
algae;
turbidity;
disinfectant decay;
groundwater level.

Do Not Compare Winter and Summer Blindly

A result that differs from last month may still be normal for the current season.

Use Related Parameters

Strong troubleshooting connects several measurements that should logically respond together.

Examples include:

raw turbidity and coagulant demand;
temperature and disinfectant decay;
ammonia and chlorine demand;
flow and chemical feed;
water age and residual;
iron and color.

Related Parameters Help Confirm a Real Change

If one parameter changes but all related variables remain stable, verify the measurement before assuming the process changed.

Correlation Does Not Prove Cause

Two parameters changing together can suggest a relationship, but the operator should still consider alternative explanations.

Determine Where the Change Begins

Compare sampling locations to identify where the abnormal condition first appears.

Possible locations include:

raw water;
treatment process;
finished water;
storage;
distribution system.

Raw Water Changes First

If raw-water parameters change before treatment results change, the source is likely contributing to the problem.

Raw Water Stable, Process Water Changes

If source conditions remain stable but treatment-stage data deteriorate, investigate:

chemical feed;
mixing;
clarification;
filtration;
equipment operation.

Finished Water Stable, Distribution Changes

If finished water remains stable but remote distribution water deteriorates, investigate:

water age;
storage;
pressure;
local hydraulics;
sediment;
main breaks.

Verify the Measurement

Before changing treatment significantly, verify unexpected data.

Possible checks include:

repeat test;
second instrument;
laboratory confirmation;
instrument calibration;
sample-point inspection.

Instrument Error

Instrument problems can result from:

dirty sensors;
expired reagents;
poor calibration;
sample-line blockage;
electrical problems;
incorrect scaling.

Sampling Error

A sample can be unrepresentative because of:

wrong location;
stagnant sample line;
poor flushing;
contaminated container;
incorrect preservation.

Data-Entry Error

Before investigating an extreme value, confirm:

decimal position;
units;
sample ID;
date and time.

Units Matter

Confusing units can create major interpretation errors.

Examples include:

mg/L versus µg/L;
gpm versus MGD;
°C versus °F;
percent versus decimal fraction.

Process Lag

A change at one location may not appear downstream immediately.

Lag can include:

hydraulic travel time;
tank detention;
reaction time;
sample-line travel time.

Do Not Compare the Wrong Time Periods

If raw water changed at 8:00 AM but finished water has several hours of detention, an 8:15 AM finished-water sample may still represent earlier source conditions.

Detention Time Estimate

A simplified relationship is:

Detention Time = Volume ÷ Flow

Detention Example

A basin contains 1 million gallons and flow is 2 MGD.

Detention Time = 1 MG ÷ 2 MGD

Detention Time = 0.5 day

0.5 day × 24 hr/day = 12 hours

The theoretical detention time is approximately 12 hours.

Theoretical Detention Is Not Exact Travel Time

Real systems can have:

short-circuiting;
dead zones;
mixing;
variable flow.

Flow Changes Affect Water Quality

Higher or lower flow can change:

detention time;
chemical dose;
filter loading;
storage turnover;
distribution water age.

Normalize Data When Useful

Total chemical use alone may be misleading when flow changes.

Useful normalized measures include:

mg/L dose;
lb per million gallons;
chemical use per unit flow.

Source-Water Troubleshooting

If source water changes unexpectedly, review:

weather;
watershed conditions;
source blending;
well operation;
reservoir conditions;
upstream activity.

Example: Raw Turbidity Increase

Review:

recent rainfall;
river flow;
reservoir turnover;
intake condition;
instrument verification.

Example: Raw Conductivity Increase

Review:

source switch;
groundwater contribution;
road salt;
industrial discharge;
instrument condition.

Example: Raw Iron Increase

Review:

well operation;
reservoir intake depth;
dissolved oxygen;
source blending.

Treatment-Process Troubleshooting

If source water remains stable but process performance changes, review:

chemical feed;
chemical strength;
mixing;
equipment condition;
process hydraulics;
instrumentation.

Example: Settled-Water Turbidity Rises

Review:

coagulant dose;
coagulation pH;
rapid mixing;
flocculation;
clarifier loading.

Example: Filter Turbidity Rises

Review:

settled-water quality;
filter run time;
filter loading;
media condition;
backwash history;
turbidimeter verification.

Example: Chlorine Residual Falls

Review:

chlorine dose;
flow;
chemical strength;
organic matter;
ammonia where relevant;
analyzer accuracy.

Distribution-System Troubleshooting

If finished water is stable but distribution quality changes, review:

tank turnover;
water age;
pressure;
flow direction;
valve operation;
main breaks;
flushing history.

Example: Remote Residual Declines

Review:

finished-water residual;
storage operation;
temperature;
water age;
local demand;
measurement accuracy.

Example: Brown Water Complaints

Review:

hydrant activity;
flow reversal;
valve operation;
iron deposits;
recent main breaks.

Example: Black Particles

Possible causes include:

manganese deposits;
premise plumbing;
other distribution deposits.

Map the Problem

Geographic patterns can provide important clues.

Map:

customer complaints;
low residual;
turbidity;
microbiological results;
pressure events.

Localized Versus Systemwide Problems

A localized problem may indicate:

one tank;
one pressure zone;
one main;
premise plumbing.

A systemwide problem more strongly suggests:

source;
treatment;
systemwide hydraulic change.

Use Independent Evidence

Confidence in a diagnosis increases when multiple independent observations support the same explanation.

For example:

rainfall;
raw turbidity increase;
higher coagulant demand;
higher sludge production.

Together, these support a real storm-related source event.

Outliers

An outlier is a result that differs greatly from surrounding data.

An outlier may be:

a real event;
sampling error;
analytical error;
instrument failure.

Do Not Delete Outliers Automatically

Investigate why the value is unusual before deciding how to interpret it.

Missing Data

Missing data should not be treated as normal results.

Operators should distinguish:

not sampled;
instrument unavailable;
result pending;
true zero.

Non-Detect Results

A non-detect means the parameter was not detected above the method's reporting capability.

It does not always mean the concentration is exactly zero.

Instrument Resolution

Small apparent changes may be insignificant if they are near the instrument's measurement resolution or normal variability.

Control Charts

A control chart can display:

normal process center;
expected variation;
unusual trends.

Control charts can help identify a process drifting before a serious limit is reached.

Operational Limits

An operational target is often designed to provide early warning.

It may be tighter than an external regulatory or permit requirement.

Do Not Wait for a Regulatory Limit to Be Exceeded

A worsening trend should be investigated before the process reaches an unacceptable condition.

Before-and-After Comparison

When an operational change is made, compare:

conditions before the change;
conditions after sufficient response time.

Change One Major Variable at a Time When Practical

If several settings are changed simultaneously, it becomes difficult to determine which adjustment caused the response.

Document Operating Changes

Useful records include:

time of change;
old setting;
new setting;
reason;
expected response;
actual response.

Short-Term Versus Long-Term Trends

Short-term trends can reveal:

equipment failure;
rapid source change;
hydraulic event.

Long-term trends can reveal:

seasonal pattern;
gradual fouling;
source deterioration;
changing chemical demand.

Operational Significance

Not every change requires immediate action.

Operators should consider:

magnitude;
rate of change;
related parameters;
risk to treatment;
proximity to operational or regulatory limits.

Prioritize Risk

Changes with potential microbiological or treatment-barrier significance generally deserve rapid attention.

Examples include:

loss of disinfection;
filter breakthrough;
pressure loss;
unexpected positive microbiological result.

Example: Raw Turbidity Up, Filter Turbidity Stable

This pattern suggests treatment is currently handling the increased particle load.

Operators should continue monitoring because treatment margin may still be reduced.

Example: Raw Turbidity Stable, Filter Turbidity Up

This points more strongly toward:

filter problem;
coagulation problem;
instrument problem.

Example: Residual Down and Temperature Up

This pattern can support the possibility of faster disinfectant decay.

Also review:

water age;
organic demand;
feed rate.

Example: Residual Down but Feed Up

Possible causes include:

increased disinfectant demand;
weak chemical;
feed-system calibration error;
analyzer error.

Example: pH Changes but Alkalinity Stable

Review:

chemical feed;
source-water pH;
instrument calibration.

Example: pH and Alkalinity Both Decline

This pattern may indicate a broader chemistry change or acid-producing process rather than only pH-instrument drift.

Example: Conductivity Changes Across the Whole System

This can suggest:

source change;
source blending change;
finished-water chemistry change.

Example: Conductivity Changes at One Location Only

Investigate:

sample location;
instrument;
local plumbing;
localized hydraulic conditions.

Use Trend Graphs Carefully

Graph scale can change how severe a trend appears.

Always check:

axis range;
units;
time interval;
missing points.

Automation Does Not Replace Operator Judgment

SCADA and online analyzers provide valuable continuous data, but operators should still verify:

instrument condition;
field observations;
laboratory results;
process response.

Alarm Review

An alarm should be treated as a signal to investigate, not simply a message to acknowledge.

Review:

alarm value;
related parameters;
equipment condition;
process effect.

Repeated Alarms

Repeated alarms may indicate:

poor alarm settings;
unstable process;
intermittent equipment failure;
an unresolved underlying problem.

Common Troubleshooting Mistakes

Reacting to one data point without verification.
Ignoring seasonal baseline conditions.
Ignoring rate of change.
Changing treatment before confirming the instrument.
Looking at one parameter without related data.
Ignoring process lag.
Comparing measurements from mismatched time periods.
Ignoring flow changes.
Ignoring sample location.
Assuming correlation proves cause.
Ignoring geographic patterns in distribution data.
Making several major process changes at once.

A Practical Unexpected-Result Review

Confirm units, time, and sample location.
Repeat or verify the measurement.
Check instrument calibration and sample conditions.
Compare the result with historical baseline.
Review related parameters.
Determine where the change first appears.
Account for process lag.
Identify likely source, treatment, or distribution causes.
Make a controlled corrective action.
Verify the response.

A Practical Trend Review

Plot the parameter over an appropriate time period.
Identify direction and rate of change.
Review minimum, maximum, and average values.
Compare with seasonal baseline.
Compare related parameters.
Review operating changes during the same period.
Identify whether the trend is operationally significant.
Document conclusions and actions.

A Practical Source-versus-Process Review

Review raw-water trends.
Review process-stage data.
Review finished-water data.
Account for hydraulic travel time.
If raw water changes first, investigate the source.
If raw water is stable but process data change, investigate treatment.
If finished water is stable but distribution data change, investigate distribution conditions.

A Practical Water-Quality Troubleshooting Sequence

Verify the data.
Define the baseline.
Identify when and where the change began.
Review related parameters.
Review flow and hydraulic conditions.
Account for process lag.
Review equipment and chemical feed.
Review source or distribution events.
Correct the most likely confirmed cause.
Trend the response after correction.

What to Remember for the Exam

Water-quality troubleshooting should begin with verification of the measurement.
A baseline is needed to determine whether a result is truly unusual.
One data point is not a trend.
Trend direction and rate of change provide clues about the cause.
Percent Change = (New - Old) ÷ Old × 100.
Averages can hide important minimum and maximum values.
Related parameters should be interpreted together.
Correlation can suggest a relationship but does not prove causation.
Determine where the abnormal condition first appears: source, treatment, finished water, storage, or distribution.
Unexpected measurements should be checked for sampling, calibration, unit, and data-entry errors.
Process lag must be considered when comparing upstream and downstream data.
Detention Time = Volume ÷ Flow is a useful estimate of theoretical process delay.
Flow changes can alter detention time, dose, loading, storage turnover, and water age.
Source-water problems should be evaluated with weather, watershed, source, well, and reservoir information.
Treatment problems should be evaluated using chemical feed, equipment, process hydraulics, and related process measurements.
Distribution problems should be evaluated using residual, pressure, water age, storage, valves, flow direction, and complaints.
Outliers should be investigated rather than automatically discarded.
Operational targets can provide warning before a regulatory or permit limit is reached.
Controlled changes and good documentation make troubleshooting more reliable.
Good water-quality troubleshooting combines verified data, trends, related parameters, hydraulics, process knowledge, timing, and operator observation.