A water bill can show the recorded or estimated volume of supplied water for the billing period. On its own, it rarely explains how much went to production, cleaning, cooling, a boiler, the product itself, or a difference that has not yet been reconciled. Start by choosing a physical boundary and time window. Then state how you will treat readings, estimates, and unknown flows.
An industrial water balance is a quantified map of inputs, uses, and outputs within a stated boundary and period. A difference between input and accounted uses is a question to investigate, not proof of a leak. The U.S. Environmental Protection Agency (EPA) describes a balance as a table, diagram, or map of flows into and out of processes. It is technical guidance, not a legal obligation for plants in Spain (EPA, Lean & Water Toolkit, Chapter 3).
What an industrial water balance can and cannot tell you
A balance answers an accounting and operational question: within the boundary we chose, what volumes entered, how were they assigned to uses, and which outputs or storage changes can we document over the same interval? The word “balance” does not mean every branch needs its own meter. It does mean that readers can tell a meter reading from an estimate, and both from a branch that has been left out.
The EPA proposes mapping flows between source, process, use, and output, then comparing measured or estimated uses with the volume recorded at the inlet meter (EPA, Chapter 3).
It helps to separate four ideas that often get blurred in conversation. Supply is the water crossing the agreed inlet point. Use describes where water goes inside the facility. Depending on the definition adopted, consumption can mean water that does not return to the same basin or water incorporated into a product, so a plant must state what its indicator means. A discharge is a flow leaving through an identified outlet.
You cannot automatically calculate discharged volume by subtracting the bill: water may evaporate, enter the product, leave through other outlets, or remain in storage. WaterSense recommends considering each supply source separately and notes that some water may not return to the sanitary sewer, for example through evaporation or incorporation into a product (EPA WaterSense at Work, §2.1).
A first balance does not replace a full audit, a water-footprint assessment, or a compliance review. It can help decide where more data are needed. By itself, it does not establish whether water is suitable for reuse, whether a discharge permit is being met, or what effect a site has on a river basin. Those questions have different boundaries, methods, and owners.
A study of a beverage manufacturer in China, for example, proposed combining accounting and mass-balance analysis to observe flows at process scale. The case shows one way to organize an analysis; it does not establish a result that can be transferred to every Spanish plant (Lyu et al., “A novel framework for water accounting and auditing for efficient management of industrial water use,” Journal of Cleaner Production, volume 395, article 136458, 2023).
The first balance is useful because it makes the state of knowledge visible. One branch may have a dedicated meter; another may be estimated from a recipe and cycle count; a third may be omitted because no record exists. This classification prevents a total from appearing more precise than its inputs. It helps decide whether available records answer the question or whether missing detail calls for another measurement.
The EPA presents this type of balance as an initial assessment of flows, comparing measured or estimated uses with recorded input (EPA, Lean & Water Toolkit, Chapter 3).
Draw the boundary and gather records before adding meters
Start with a boundary you can describe in one sentence. It might cover an entire plant from the inlet meter to identified outputs, or one production line from the point where it receives water to its drain and known transfers. Record which tanks fall inside it, which connections cross it, whether there is an alternative supply, and where water can accumulate between periods. A simple, legible sketch is often enough for a first pass.
If it does not show the actual route, correct it with people who know the site before using it to allocate volumes. The EPA allows a table, diagram, or map to represent flows, depending on the facility (EPA, Chapter 3).
Next, identify inputs by origin: mains water, a well, reclaimed water, or another source present at the site. Do not assume the bill is the only input. WaterSense recommends accounting for water arriving from each source and measuring each source separately where appropriate (EPA WaterSense at Work, §2.1). That guide addresses commercial and institutional buildings in the United States. Here it is a reference for measurement detail, not billing advice for a Spanish company.
Check meters, delivery points, and supplier documentation with the responsible team.
Then walk through the relevant uses: water incorporated into a product or ingredient, process operations, manual or automated cleaning, CIP where present, cooling towers or circuits, boilers, auxiliary services, and sanitary uses. For each point, note where the data came from, its unit, the period it covers, and who can explain its meaning. If a drawing says a line supplies a cooling tower but an actual branch also feeds a cleaning operation, record that difference.
An asset name does not replace a physical check.
First gather what already exists: bills and readings, piping plans or diagrams, production records, cleaning recipes or cycles, work orders, operating data, documented cooling-tower and boiler controlled-discharge volumes, and storage changes. The guide to preserving context in industrial data explains why the origin, period, and scope of each record matter when preparing a decision. For this work, a reading with no clear date or a production figure with no unit is a clue, not yet a comparable basis.
An inventory sheet can organize the review. It might include the point or flow, origin and destination, unit, instrument or method, frequency, period covered, status (measured, estimated, unknown, or excluded), supporting record, and notes. There is no need to fill gaps with generic percentages. If an important flow has not been quantified, the balance can show it as pending and name the observation that would reduce uncertainty.
A study of audits at two Australian manufacturers describes a sequence of flow characterization and balance diagrams as a basis for later analysis. Its scope is those two cases, not a guarantee of results in other sectors (Agana et al., “An approach to industrial water conservation – A case study involving two large manufacturing companies based in Australia,” Journal of Environmental Management, volume 114, pages 445–460, 2013).
Lyu and co-authors describe process-scale water accounting integrated with mass balance at another specific company (Lyu et al., “A novel framework for water accounting and auditing for efficient management of industrial water use,” Journal of Cleaner Production, 2023).
Choose periods and meters that support a fair comparison
A totalizer and an instantaneous flow reading answer different questions. A totalizer accumulates volume between two readings. Instantaneous flow describes the rate at one moment, while average flow summarizes a rate over an interval. To calculate volume from flow readings, integrate them over time and retain the timestamps. Before combining records, confirm what each column represents and which unit it uses.
WaterSense recommends understanding the capabilities of existing meter infrastructure to see what resolution each reading offers (EPA WaterSense at Work, §2.1). “m³” may be accumulated volume; “m³/h” is a flow rate. You can multiply an average flow rate by a duration only when the rate represents that exact interval. A momentary reading alone cannot estimate a whole shift’s volume.
Align dates before adding values. A monthly bill, a reading at the start of each shift, and production for a campaign may use different calendars. WaterSense distinguishes conventional meters, which are commonly read monthly or quarterly, from advanced meters that may record at least hourly and transmit more often. This describes general capabilities; it is not a specification for buying a particular meter (EPA WaterSense at Work, §2.1).
A monthly meter reading can support the volume for that interval; it does not show how use was distributed across shifts.
Before adding separate meters, ask what decision would change with the new data. If you need to know whether one cleaning event explains a variation, a meter on a branch shared by the whole plant may not be enough. To distinguish cycles on a line, the measurement point must capture the relevant flow, and the logging window must fit the cycle. There is no universal meter model here.
Selection requires reviewing the fluid, operating range, installation, materials, process conditions, and maintenance requirements with competent staff and the manufacturer’s documentation. WaterSense also distinguishes understanding installed infrastructure from deciding which new capabilities are needed; its guide does not specify a meter for a particular plant (EPA WaterSense at Work, §2.1).
Also check what often disappears from a spreadsheet: jumps or resets in the cumulative-volume meter register, clock changes, time zones, missing periods, instrument replacements, manual readings, and unit conversions. Keep the original data beside any transformation. If you estimate a missing interval, mark the method and period as an estimate. Do not fill gaps without leaving a record.
The Guide to the Expression of Uncertainty in Measurement, issued by the Joint Committee for Guides in Metrology and the International Bureau of Weights and Measures (GUM) brings together uncertainty components evaluated from observations and other information; combined uncertainty depends on the measurement model and its input quantities. The guide gives general rules, not instructions for one technology or a universal tolerance for water meters (GUM 100:2008).
If you compare water per tonne of production, record the production basis, product or mix, and exact window too. Comparing periods with different campaigns may describe two situations, but it does not show that one caused the other. The guide to comparing plant indicators measured in different ways helps check that definitions, context, and timing match before you interpret the ratio.
Separate process, cleaning, and utility uses without mixing their cycles
Trace actual operations when assigning process water; bill categories alone rarely show where the water goes. Draw branches from the source to equipment, stages, or areas and mark transfers between them. When water is used as an ingredient or remains in a product, record that destination under the site’s chosen definition. When water passes through a process and leaves through a drain, record the route without assuming the outlet is at the inlet meter.
If several lines share a supply, do not split the total equally without a documented method. A published framework for industrial water accounting and auditing also proposes tracing water through production links, though it illustrates its method with one beverage company (Lyu et al., “A novel framework for water accounting and auditing for efficient management of industrial water use,” Journal of Cleaner Production, 2023).
Cleaning deserves its own view because water use may occur in cycles and phases, while other uses are continuous or depend on load. If the plant uses CIP, start with a real cycle identified in the records: pre-rinse, wash, rinses, recovery, return, and discharge, but only where those phases exist. Connect each phase with the line or equipment, product or campaign, date, cycle count, measured volume or estimation rule, and water destination.
The European reference document on best available techniques for the food, drink, and milk industries records examples of measurement and transition control in CIP systems (European Commission, FDM reference document, Chapter 2). If there is no meter for each phase, an estimate based on preset volume and event count may be possible, but label it as an estimate and preserve its source.
The European best available techniques reference document for the food, drink, and milk industries describes examples of CIP systems in that sector where conductivity meters help control phase changes, and flow meters with preset volumes can support control of water use. The document also limits applicability according to the equipment and whether it can be adapted.
These are techniques described in a sector context, not a recommendation to shorten cycles or a promise of savings in another industry (European Commission, FDM reference document, Chapter 2, p. 117). Any change to a recipe or sequence requires the plant’s assessment and approval process, including hygienic validation where applicable.
Do not forget utilities. A cooling tower may have recorded make-up, controlled discharge, and evaporation. A boiler may have water supplied to it, controlled discharge, and recovered condensate. Represent these branches according to the actual configuration and available data. The EPA guide includes evaporation and wastewater discharge among the flows a balance may consider, although the exact categories depend on the process and chosen boundary (EPA, Chapter 3).
WaterSense also notes that evaporated water or water incorporated into a product may not return through the sanitary sewer (EPA WaterSense at Work, §2.1). Do not treat a design rate for controlled water discharge as a meter reading or assume condensate always returns to the same point.
Finally, separate reuse from transfer. Water moving from one stage to another inside the boundary is not a new external input, though it can be useful to record as an internal flow to understand uses. If a stream is reused, identify its origin, treatment or condition as documented by the facility, destination, and known quantity. This balance does not determine whether its quality is suitable for a particular use.
Reconcile inputs and outputs: a difference does not mean a leak
For the plant boundary, use this convention: unreconciled difference = external inputs − external outputs − change in storage, where change in storage = ending volume − starting volume. A positive difference means recorded inputs exceed the sum of accounted outputs and the increase in storage; a negative one means the reverse. External outputs are flows that cross the chosen boundary, such as discharges or water leaving in the product.
Process uses help allocate and understand water inside the plant. Do not add them again as external outputs if they are already represented there. Draw internal transfers, but do not count them as a new input or output at the boundary. Write the convention beside the calculation and note whether a tank held a different volume at the start and end.
Audit studies that pair balance diagrams with flow tests offer examples of analysis within defined boundaries; they do not set one universal model for every plant (Agana et al., “An approach to industrial water conservation – A case study involving two large manufacturing companies based in Australia,” Journal of Environmental Management, 2013).
Classify every item in the sum as measured, estimated, included but unquantified, or excluded. Keep original values, conversions, dates, and data owners. Then check that the windows match. A thirty-day bill should not be compared directly with readings from branch meters covering a nominal thirty days if they began or ended at different times, unless you document the adjustment and its limits.
WaterSense stresses the value of understanding where readings come from, how often they are recorded, and which supply sources they cover (EPA WaterSense at Work, §2.1).
Express the difference first as a volume over a period, not as a diagnosis. A percentage adds context only when the denominator is valid and the quantities share a boundary and time window. If a large branch is estimated or a period is incomplete, calculate the percentage with that limitation in view, or leave it out. The international GUM provides general rules for evaluating and expressing uncertainty.
The calculation depends on the model and input quantities; it does not set a tolerance that applies to every water meter (GUM 100:2008, sections 1, 2.2 and 2.3).
An unreconciled difference does not locate its cause. It may reflect uses that were not included, reading dates that do not line up, inventory changes, a reset totalizer, a unit conversion error, an unknown outlet, evaporation, or a physical loss. The EPA presents the balance as an initial assessment and a starting point for directing an investigation; that role does not mean the difference is a leak (EPA, Chapter 3).
If a leak is a serious possibility, you need additional, focused evidence.
Before requesting a new installation, review the data with the greatest potential effect: inlet reading and date, known branches, cleaning cycles, missing periods, and storage conditions. Ask operations and maintenance about starts, stops, purge events, or work that coincided with the window. The guide to energy losses and plant evidence makes a caution that applies here too: an unusual total can show where to look, but it does not establish a causal mechanism.
Turn the first balance into a verifiable investigation
The first useful result may be a map that marks each branch as measured, estimated, or unknown, plus a list of checks still needed. The figures do not have to reconcile perfectly. The next step should answer a specific uncertainty. If the biggest unknown is cleaning water per cycle, identify which meter, recipe record, or authorized test could reduce it. If dates are the problem, fix the calendar first.
Installing instruments without knowing which question they should answer adds data, but not necessarily knowledge.
Prioritize hypotheses by their relevance to a decision and by the quality of the evidence. A variation that coincides with a campaign or production change should be compared with a similar period before it is attributed to a loss. A reading that changes just after a meter replacement may call for a review of the equipment change and its records. An undocumented route may justify an approved visual inspection. Keep a clue, a hypothesis, and a confirmed cause distinct.
The responsible person should verify findings under local procedures. The guide to maintenance investigations based on plant evidence explains the difference between an indication and a conclusion.
A short investigation record can include the question, boundary, period, observed result, sources used, assumptions, limitations, hypotheses, proposed check, owner, and review date. When you close a hypothesis, keep the result even if it does not confirm the first explanation. A later reading can then improve the balance without erasing why the team made its earlier decision. If available data are not enough, write down what is missing instead of filling the gap with a generic value.
To compare periods, group days or shifts with operating conditions that make sense for the question. Note product, production volume, calendar, planned shutdowns, and relevant cleaning events when those data are available. This does not turn an association into causality. It avoids comparing windows that represent different kinds of operation.
WaterSense explains that conventional readings are often monthly or quarterly, while advanced meters may record more frequently; the detail available limits which periods can be compared (EPA WaterSense at Work, §2.1).
Published research on process-scale water accounting and balances explains why observing flows alongside production activity can help characterize an industrial case, but it reports a study at one company, not a universal rule (Lyu et al., “A novel framework for water accounting and auditing for efficient management of industrial water use,” Journal of Cleaner Production, 2023).
Once you have enough evidence to answer the question, document what changed in your understanding of the flow and which decision sits outside the balance’s scope. A team can use its records and site context to prepare a traceable review. WizeeMind describes its role as a knowledge assistant and decision-support tool with human approval. This article makes no claim that the product connects to meters, controls CIP cycles, or detects leaks automatically.
The people responsible for the facility should decide what to check next under its operating procedures.
FAQ
When can a water balance be considered closed?
Document the calculation when its inputs and outputs share a boundary and period, each volume is marked as measured or estimated, and assumptions and exclusions are recorded. The EPA says to verify that flows in and out balance; it also describes using root-cause analysis to investigate water waste (EPA, Lean & Water Toolkit, Chapter 3). A balance is reconciled only when its accounted flows balance within the stated boundary and period and any discrepancies have been investigated and explained.
If a difference remains unexplained, describe it as unresolved: the calculation may be documented, but the balance is not closed. To interpret the difference, use comparable data and an account of uncertainty based on the measurement model (GUM 100:2008, sections 1, 2.2 and 2.3). Keep the boundary and dates with the result so the next review can see what was included and compare it on the same basis.
References: EPA; GUM 100:2008.
Can I compare a monthly bill with daily or shift readings?
You can compare the bill total with data aggregated over the same window if the measurement point, unit, and start and end dates match. A monthly reading cannot allocate volume among days or shifts without other evidence and an explicit allocation rule. Check for estimated readings or gaps before interpreting the difference; available resolution limits the time detail you can compare (EPA WaterSense at Work, §2.1).
Keep both readings and their dates so someone else can repeat the calculation without assigning monthly consumption to shifts without evidence.
References: WaterSense at Work.
How should I include cleaning water or a CIP cycle?
Record the phases that actually take place, the date, equipment, cycle count, volume, and destination. If a reading is missing, write down the estimation method and do not present the estimate as a measurement. The European reference document for the food, drink, and milk sector describes examples of CIP phase control, but its scope is sector-specific (European Commission, FDM reference document, Chapter 2, p. 117).
Group cycles by equipment or campaign where the records allow, and keep measured values separate from estimates. Do not change a cleaning recipe or sequence based only on the balance.
References: FDM document.
Does the balance difference prove there is a leak?
No. It may reflect an unmeasured use, misaligned periods, storage changes, evaporation, or recording and measurement errors. The EPA uses the balance as an initial assessment; the difference alone does not identify its cause (EPA, Lean & Water Toolkit, Chapter 3). Treat it as pending reconciliation and test any explanation with additional evidence. If you combine measurements with different quality or frequency, first describe the model and its inputs.
The GUM provides general uncertainty rules, not a universal tolerance for water meters (GUM 100:2008).
References: EPA; GUM 100:2008.