A wet-weather flow increase does not name its own source
After rain, a sewer carries more water. That observation matters, but it does not yet tell us whether the extra water arrived through a surface connection, travelled through the ground, displaced stored water or accompanied a change in ordinary wastewater use. Sewer exchange becomes clearer when we separate the pathways before calculating their size.
Groundwater infiltration is water entering a sewer from its surroundings. Exfiltration is water leaving it. Surface inflow is a different route into the system. This guide develops the signs, volumes, tracer balances and timing needed to distinguish these possibilities without calling every unexplained difference a leak.
The examples use supplied fictional records. They are not instructions to enter a sewer, open infrastructure or release a tracer. Real investigations require the responsible operator, qualified practitioners, lawful access and appropriate environmental and safety controls. The useful learning task here is to audit the inference.
Hydrology subject guide · Learning that transfers into life · How X Works hub
Choose your reading route
Route 1: Define direction and boundary · Chapters 1–3
Route 2: Build the balances · Chapters 4–5
Calculate net exchange without pretending it is gross exchange
Route 3: Test tracer evidence · Chapters 6–7
Route 4: Read rain and operational timing · Chapters 8–10
Route 5: Transfer and explain · Chapters 11–12
Full chapter index · Independent worked case · Questions and practical checks
Chapter index
Define direction and boundary · Chapters 1–3
Keep three pathways separate
Use hydraulic head, not a simple above-and-below picture
Choose a control volume that can be audited
Build the balances · Chapters 4–5
Calculate net exchange without pretending it is gross exchange
Keep storage and travel time in the comparison
Test tracer evidence · Chapters 6–7
Derive a tracer mixture before trusting its fraction
Test endmember contrast, background and missing sources
Read rain and operational timing · Chapters 8–10
Separate concentration change from transported marker mass
Read wet-weather response as a hypothesis test
Locate exchange without overextending a network result
Transfer and explain · Chapters 11–12
Build independent checks around the weak link
Write a bounded conclusion that supports a next step
CHAPTER 1 OF 12
1. Keep three pathways separate
Start with a drawing containing a pipe, the surrounding ground and a possible surface connection. An arrow from ground into pipe is infiltration. An arrow from pipe into ground is exfiltration. An arrow carrying surface water directly into the network is inflow. The labels describe direction and route, not simply whether water was unwanted.
A crack is evidence of a possible connection, not a flow meter. Water movement also depends on hydraulic conditions, surrounding material and whether the pathway is open. A visible defect can be important while supplying neither a measured leakage rate nor a constant direction. Do not multiply a photograph by the number of similar-looking pipes and call the result a network water budget.
The practical question should name the pathway: estimate groundwater entering a defined network during a stated period, or test whether wastewater is reaching a particular surrounding-water receptor. These are different investigations. Evidence adequate to explain increased treatment-plant loading may still be inadequate to locate an environmental release.
CHAPTER 2 OF 12
2. Use hydraulic head, not a simple above-and-below picture
Water can move through a connected defect when the relevant hydraulic head differs across it. Compare inside and outside conditions on a common elevation reference. Pipe-water depth alone is not enough if the surrounding groundwater level is unknown. Likewise, a regional groundwater contour cannot automatically describe the local head beside a particular pipe.
Consider a fictional defect at elevation 10.0 metres. The sewer water surface is at 10.5 metres and the nearby groundwater head is at 11.0 metres. The outside-to-inside head difference is +0.5 metres, supporting inward movement if a conducting connection exists. The example establishes a driving direction under its assumptions, not a rate.
Now let sewer surcharge raise its water surface to 11.4 metres while outside head stays at 11.0 metres. The difference reverses. The same defect could support outward movement instead. Actual rates require information about the connection and its response; a direction inferred for one hour should not silently become a permanent property of the pipe.
CHAPTER 3 OF 12
3. Choose a control volume that can be audited
Draw the upstream and downstream monitoring boundaries and list lateral connections, pumps and temporary storage inside them. A network is not a rigid bucket with one inlet and one outlet unless the chosen simplification genuinely fits. Unrecorded lateral wastewater and changing pipe storage can mimic an exchange signal.
Use one period and compatible units. A flow of 5 L/s maintained for an hour is 18 cubic metres: 5 × 3,600 ÷ 1,000. A daily volume cannot be subtracted directly from an instantaneous flow rate. Convert rates to volumes over the same interval before applying conservation.
State the sign convention before examining the residual. Here infiltration and ordinary wastewater are inputs, exfiltration and downstream discharge are outputs, and positive storage change means more water remains inside. If a colleague chooses a different sign convention, the physics need not differ. The equations must simply preserve the declared directions consistently.
CHAPTER 4 OF 12
4. Calculate net exchange without pretending it is gross exchange
Let Vw be known wastewater input, Vi infiltration, Ve exfiltration, Vo outlet volume and ΔS storage change. With other pathways excluded for the teaching case, ΔS = Vw + Vi − Ve − Vo. Rearranging gives Vi − Ve = Vo + ΔS − Vw. The right-hand side estimates net exchange into the sewer under those exclusions.
If wastewater input is 100 m³, outlet volume 125 m³ and storage increases by 5 m³, net inward exchange is 30 m³. Both Vi = 30 and Ve = 0, and Vi = 50 with Ve = 20, satisfy that budget. The total inward movement is not uniquely determined by the net balance.
This distinction matters when direction changes between locations or hours. An apparently small daily net exchange can coexist with substantial inward and outward flows. Dividing a net residual by pipe length does not recover gross leakage from individual defects. Report exactly which quantity the measurements identify and which additional evidence would separate the opposing terms.
CHAPTER 5 OF 12
5. Keep storage and travel time in the comparison
Suppose an upstream flow pulse enters at 10:00 and reaches a downstream monitor later. Comparing identical clock readings without accounting for travel and storage can create a temporary apparent loss followed by an apparent gain. The pipe has transported and stored water; the difference does not automatically represent exchange with the ground.
For an integrated example, an interval receives 80 m³ and releases 65 m³ while measured internal storage rises by 15 m³. The budget closes without leakage. If storage is omitted, the missing 15 m³ can be wrongly labelled exfiltration. When the stored water later leaves, the same mistake may produce an invented infiltration episode.
Pump cycling can change both timing and hydraulic conditions. Preserve pump-state records and the basis of each flow estimate. A comparison should distinguish actual boundary volume from water recirculating inside the chosen system. Increasing the number of sensors does not help if their clocks, boundaries or quantities are incompatible.
CHAPTER 6 OF 12
6. Derive a tracer mixture before trusting its fraction
A tracer is useful when it distinguishes candidate water sources and behaves suitably over the comparison. Consider a conservative marker with wastewater concentration Cw, groundwater concentration Cg and mixed concentration Cm. For two fully mixed endmembers, Cm = (1 − f)Cw + fCg, where f is the groundwater fraction of the mixed water.
Rearranging gives f = (Cw − Cm) ÷ (Cw − Cg). If Cw = 12, Cg = 2 and Cm = 9 in one consistent concentration unit, f = 3/10 = 0.30. For 100 m³ of mixture, that means 30 m³ assigned to the groundwater endmember under this model. It is not a measured pipe-defect discharge.
The assumptions carry the interpretation: representative endmembers, two-source mixing, compatible timing and negligible marker change along the route. Zhao and colleagues used a chemical tracer and a physically based model to investigate groundwater entering an urban sewer. Their site-specific research supports the value of combined evidence, not a universal concentration or leakage percentage.
CHAPTER 7 OF 12
7. Test endmember contrast, background and missing sources
The denominator Cw − Cg is the tracer contrast. If wastewater and groundwater have nearly identical concentrations, a small measurement or endmember error can produce a large change in f. A precise laboratory number does not solve a source-separation problem when the sources themselves overlap.
Change the previous example to Cw = 10, Cg = 9 and Cm = 9.7. The nominal fraction is again 0.30. But changing Cm to 9.6 gives 0.40, while 9.8 gives 0.20. A change of only 0.1 concentration units now moves the inferred fraction by ten percentage points. Compare that sensitivity with the earlier ten-unit endmember separation.
A third input can break the two-source interpretation. Low-marker surface inflow may dilute wastewater in much the same way as groundwater. A concentration decrease therefore needs route evidence, timing or additional source information before it is assigned specifically to groundwater. Values outside the endmember range are a diagnostic warning, not a reason to clip fractions quietly to zero or one.
CHAPTER 8 OF 12
8. Separate concentration change from transported marker mass
A concentration can fall while marker load remains unchanged. Suppose 100 m³ of wastewater at 10 mg/L mixes with 100 m³ of marker-free water. Initial marker mass is 1 kg because 100,000 litres multiplied by 10 mg/L equals 1,000,000 mg. The final 200 m³ at 5 mg/L still contains 1 kg.
The dilution has doubled water volume, not removed marker mass. Calling the 50% concentration reduction a 50% pollutant removal would be incorrect for this fictional conservative case. Conversely, a marker mass deficit could reflect an outward pathway, reaction, sampling mismatch or an omitted boundary, depending on the system.
Compute load from concentration and matching flow through time, not from a single grab sample multiplied by any convenient daily total. If concentration and flow covary, their average product differs from the product of separate averages. Preserve the underlying intervals and explain how missing observations affect the integrated mass estimate.
CHAPTER 9 OF 12
9. Read wet-weather response as a hypothesis test
A rapid rise during rain is consistent with a fast connection, but the shape alone does not uniquely establish a route. A delayed tail can reflect groundwater response, network storage, continued surface drainage or operational changes. Compare candidate explanations with rainfall, outside heads, pump records and spatially distributed evidence.
A dry-weather baseline is a model of expected wastewater under stated conditions. It can vary by hour, day type, occupancy and industrial activity. Subtracting a fixed baseline from every wet-weather reading attributes all departures to rain-related water, including departures caused by changed wastewater use. The residual inherits that baseline uncertainty.
Use matched comparisons where possible and preserve alternatives. If a wet Tuesday is compared with a dry public holiday, ordinary demand may differ. If a pump regime changes between periods, timing may differ. A useful report says which comparison was used, why it is credible, and which departures it cannot isolate.
CHAPTER 10 OF 12
10. Locate exchange without overextending a network result
An outlet mixture can constrain a network total while leaving location unresolved. Two upstream areas can produce the same combined tracer and flow record. To distinguish them, obtain spatial evidence suited to the question rather than distributing the outlet estimate evenly along every pipe.
Suppose a 20 m³ net inward daily residual comes from either one short reach or several scattered reaches. Both explanations fit the aggregate number. Inspection may identify candidate connections, and authorised local monitoring can test them. Until then, a map showing equal leakage intensity everywhere is a mapping assumption, not an observation.
Environmental attribution needs another link: source, pathway and receptor. Finding a wastewater-associated marker in surrounding water can motivate investigation, but its presence does not by itself identify a particular pipe or release time. Backgrounds, other sources, travel and transformation require attention. Do not turn a network accounting exercise into a public-health or legal conclusion it cannot support.
CHAPTER 11 OF 12
11. Build independent checks around the weak link
If direction is uncertain, improve matched head evidence. If net volume is uncertain, inspect flow, storage and lateral inputs. If source fraction is uncertain, test endmember contrast and alternate sources. If location is uncertain, seek spatially discriminating observations. This is more effective than collecting a larger volume of the same non-discriminating data.
For a teaching exercise, hide one piece of information and ask what conclusion remains possible. Remove the groundwater concentration: the two-source fraction is no longer computable. Restore it but remove storage change: a network volume residual becomes ambiguous. Restore both but add an unmeasured third source: the apparently complete calculation needs a revised model.
Separate fit from evaluation when using an optimisation model. A model adjusted to reproduce one outlet series has not automatically identified the true spatial exchange pattern. Check against evidence not used to tune it, and report alternative parameter arrangements that fit similarly. Repeating a good fit is not the same as independently resolving the mechanism.
CHAPTER 12 OF 12
12. Write a bounded conclusion that supports a next step
A useful conclusion identifies quantity, boundary, period and uncertainty. For example: the monitored reach showed net inward exchange over the observation day under the stated wastewater and storage estimates; tracer evidence supported a low-marker additional input, but groundwater and surface-inflow contributions were not separately resolved. The limitations are specific enough to guide the next investigation.
Retain the signed budget, source-concentration evidence, time alignment, missing data and operational context. Distinguish measured terms, estimated terms and residuals. State whether the result is instantaneous flow, interval volume, net exchange, gross exchange or source fraction. These labels prevent a technically correct number from answering the wrong question.
The transfer goal is not memorising a formula. It is recognising when a formula still applies after the system changes. Reverse the heads, add storage, introduce a third source or reduce tracer contrast. Explain which conclusion survives each change and which extra observation is needed. That is the reasoning a real water claim deserves.
Independent worked case and evidence checks
Consider a fictional monitored reach over six hours. Known wastewater input is 180 m³. Outlet volume is 216 m³. Internal storage rises by 12 m³. Assume for the first calculation that surface inflow, unrecorded laterals and other pathways are absent. All volumes refer to the same boundary and period.
The signed balance gives Vi − Ve = 216 + 12 − 180 = 48 m³ net inward exchange. Omitting storage would give 36 m³ and understate the net input by 12 m³. The stored water is inside the boundary at the end of the interval even though it has not reached the outlet.
One compatible explanation is 48 m³ infiltration and zero exfiltration. Another is 60 m³ infiltration and 12 m³ exfiltration. A third is 90 m³ infiltration and 42 m³ exfiltration. The outlet budget alone cannot choose among these gross-flow pairs.
Now change the storage observation to a decrease of 12 m³. The net inward result becomes 216 − 12 − 180 = 24 m³. The same outlet volume can support a different exchange estimate because part of the outlet water came from previously stored water. The sign of storage is a physical distinction, not a bookkeeping preference.
Next introduce independently justified two-source tracer evidence for the outlet mixture. Let wastewater concentration be 14 units, groundwater concentration 2 units and mixed outlet concentration 11 units. The inferred groundwater fraction is (14 − 11)/(14 − 2) = 0.25. Applied to 216 m³ of outlet water, this assigns 54 m³ to the groundwater endmember at that boundary.
Do not demand that 54 immediately equal the 48 m³ net-exchange calculation. One describes source composition of water leaving the boundary; the other describes net exchange over a control volume with changing storage. The source composition of initial and final storage, exfiltration and travel histories can matter to their reconciliation.
For a simpler closed two-source mixing exercise, remove storage and exfiltration and let 180 m³ of wastewater mix with 60 m³ of groundwater. The outlet is 240 m³ and groundwater fraction is 0.25. Its concentration is (180 × 14 + 60 × 2)/240 = 11 units. Now the water and marker balances refer to compatible assumptions and agree.
Keep that 240 m³ volume and 11-unit mixture but introduce 60 m³ of surface inflow also at 2 units instead of groundwater. The same marker calculation produces the same fraction. It identifies the low-marker endmember, not its route. Independent route evidence is essential if the claim specifically says groundwater infiltration.
Try a consistency test. With only 14-unit and 2-unit endmembers, an observed mixture at 16 units is outside the permitted range. The two-source model cannot explain it with fractions between zero and one. Investigate source variability, an omitted source, sampling or analytical problems before accepting a mathematically negative groundwater fraction.
Now test endmember uncertainty. If the wastewater value in the 11-unit case were 13 rather than 14, the inferred fraction would be (13 − 11)/(13 − 2) = 2/11, about 0.182. If it were 15, the fraction would be 4/13, about 0.308. Treating a variable wastewater endmember as exact can make a source estimate look more stable than it is.
Inspect time averaging with two equal-duration intervals. In the first, flow is 10 L/s and marker concentration 4 mg/L. In the second, flow is 2 L/s and concentration 12 mg/L. Mean marker transport is (10 × 4 + 2 × 12)/2 = 32 mg/s. Multiplying separate means gives 6 × 8 = 48 mg/s, which is wrong for this record.
For two one-hour intervals, integrate the matching loads: 40 mg/s × 3,600 s plus 24 mg/s × 3,600 s = 230,400 mg, or 0.2304 kg. The product-of-means shortcut would give 0.3456 kg. The discrepancy is caused by ignoring the relationship between flow and concentration, not by insufficient decimal places.
A separate head record shows outside head 20.8 metres and inside water level 20.3 metres for the first hour. Later surcharge raises the inside level to 21.1 metres while outside head remains 20.8 metres. Under a connected-pathway assumption, the head difference changes from inward-driving to outward-driving. It still does not reveal the conducting area or gross volume.
An inspection finds two defects, one in each of two reaches. It would be tempting to split the inferred exchange equally. There is no basis for that split in the supplied data. Different local heads, defect responses and surrounding soils could give very different contributions.
Finally, compare wet-weather attribution. A fixed dry baseline predicts 30 m³/h for six hours, or 180 m³. But a matched day-type analysis suggests ordinary wastewater could range from 27 to 33 m³/h. That corresponds to 162–198 m³. With the original outlet and +12 m³ storage, net additional input ranges from 30 to 66 m³ before considering other uncertainties.
This range is a sensitivity calculation, not a probabilistic confidence interval. It shows how a stated baseline alternative changes the answer. Do not label its endpoints a 95% interval unless a defensible statistical model actually supports that interpretation.
There is one more useful error check. Suppose actual inlet and outlet volumes are both 200 m³, storage is unchanged and there is no exchange. If the inlet monitor reads 5% high and the outlet monitor reads 5% low, the recorded values are 210 and 190 m³. The apparent net exchange is −20 m³. A paired measurement bias has manufactured apparent outward loss.
If both monitors instead read 5% high, they report 210 and 210 m³ and the net difference remains zero. Individual uncertainty percentages cannot be inserted blindly into a budget without considering error direction and shared behaviour. Cancellation in this special zero-difference example does not prove that all future balances will be unaffected.
Change the real volumes to 200 m³ entering and 240 m³ leaving, with unchanged storage and 40 m³ true net input. A common 5% scale bias gives 210 and 252 m³, producing 42 m³ rather than 40. Even shared errors that partly cancel can affect a nonzero difference. Keep calibration evidence and the uncertainty model connected to the quantity being estimated.
For retrieval practice, ask which new observation would repair each example: a better tracer cannot alone correct biased flow volumes; a better outlet monitor cannot alone establish the groundwater endmember; and matched head records cannot alone locate every exchanging defect. Each supplies a different missing relationship. A good investigation chooses evidence for that relationship instead of treating all additional data as interchangeable.
The final report should distinguish four statements: a measured outlet volume, a storage-adjusted net budget, an assumption-dependent tracer fraction and unresolved route or location. The next useful observation follows from which unresolved distinction matters to the decision. No one needs to enter infrastructure to learn how to ask those questions.
Worked comparison
| Evidence | Supports under stated assumptions | Does not establish alone |
|---|---|---|
| Inside and outside head | Potential driving direction at that time | Defect conductance or exchanged volume |
| Inlet, outlet and storage volumes | Net boundary exchange | Separate gross infiltration and exfiltration |
| Two distinct conservative endmembers | Mixture source fraction | Route if another source has the same signature |
| A rain-related outlet rise | A timing association | Unique groundwater or surface-inflow attribution |
Questions and practical checks
Is infiltration the same as inflow?
No. Here infiltration enters from surrounding ground, while inflow follows a surface connection. Both may increase sewer loading, but their routes and useful checks differ.
Can a sewer both gain and lose water?
Yes, different locations or periods can have different driving conditions. A net daily budget does not separately identify all inward and outward movements.
Does dilution prove that leakage stopped?
No. Concentration, water volume and transported marker mass must be distinguished. Dilution can lower concentration without removing marker mass.
Can a negative mixing fraction be accepted?
It is a warning that the assumed endmembers, observations or two-source model do not fit. Do not silently clip it and present the clipped number as an established source estimate.
What is a safe learning exercise?
Audit supplied flow, concentration, head and storage records. Real sewer access and tracer work belong to authorised operators and qualified investigations, not classroom imitation.
Continue with the connected subjects
Follow urban routes and accounting
Urban dual drainage: inlet capture, surcharge and surface routing
Water-budget closure: correlated errors and unresolved residuals
Check transport and groundwater interpretation
Tracer dilution: mixing, background and recovery checks
Groundwater recharge estimates: independent methods and uncertainty
Sources and scope
Zhao and colleagues, 2020: tracer and model evidence for groundwater entering urban sewers
Rutsch, Rieckermann and Krebs, 2006: review of sewer-leakage quantification
US EPA: sanitary sewer overflows and collection-system context
The numerical cases in this guide are original fictional teaching examples, not measurements from a real site. This is an educational explanation, not a field safety instruction, engineering approval or an official certification. For a real project, use qualified practitioners, current product documentation and site-specific evidence. Sources were checked on 30 October 2026.