“100x faster” - what it really means in structural monitoring

“100x faster” is the ratio of two intervals: how often you learn what the structure is doing with periodic measurement, and how often you learn it with continuous monitoring and readings at the default 15-minute interval. A daily reading gives 96x, a weekly one 672x, and a quarterly one 8 640x. This is arithmetic, not an outcome from a deployment, which is why you can calculate it for your own asset before you speak to anyone and decide for yourself whether the number changes anything for you.

In brief

• The speed of structural monitoring is measured by three clocks: from event to knowledge, from question to answer, and from data to report.
• Multiplier = periodic measurement interval ÷ 15 min. A quarter is 8 640x, a month 2 880x, a week 672x, a day 96x. “100x” is a cautious formulation on this scale.
• Shortening the loop changes operational decisions: stopping works that same morning instead of after the next reading.
• “Faster” will not fix bad thresholds, a missing procedure, an unattended phone, or an uncalibrated sensor. Those are separate conditions that must be met.
• At the end, there is a formula and space for your own numbers.

A 150-word manifesto

A structure does not wait for a measurement schedule. A diaphragm wall moves when the excavator removes the next layer, not when the surveyor has a free Tuesday. A hall beam deflects under snow at night, not on the inspection date. Between two periodic readings, the asset is blind, and so are you.

Continuous monitoring is not “better measurement”. It is a different kind of knowledge: instead of points on a timeline, you have a line; instead of a report after the fact, you have information in progress. That difference can be calculated. If you measure once a quarter today and the logger reads sensors at the default 15-minute interval, you learn about a change 8 640 times more often. If you measure once a day, it is 96 times. The slogan “100x faster” is, on this scale, more likely understated than overstated.

This is not a safety promise. Fast information without good thresholds, without a person who answers the SMS, and without a procedure that says what to do is only faster noise. Speed is a necessary condition. The rest is engineering work.

Three clocks used to measure structural monitoring

When we talk about monitoring speed, we usually mix three different things. Each has its own clock, and each is measured differently. Let us separate them, because without that, the dispute over “real time” is a dispute over words. If you hear “100x” and think “marketing”, that is a healthy reaction. Below is the divisor, not a slogan.

Clock 1: from event to knowledge

The “from event to knowledge” clock is the time between the moment the structure exceeds a limit value and the moment the responsible person knows about it. In a periodic model, the path looks like this:

1. Event. The exceedance happens at a random moment between readings. On average, it is halfway through the interval, and in the worst case just after the last measurement.
2. Reading. The team arrives according to schedule, after a day, week, month, or quarter. Until then, nobody knows anything.
3. Travel and measurement. Several hours on site, sometimes with a lane closure or a work stoppage during the measurement.
4. Logger. Data go to a laptop, a logger file, or a field notebook.
5. Spreadsheet. Someone transcribes or imports the readings, adds temperature compensation, calculates differences against the zero reading, and draws the chart.
6. Email. The report goes to the contract manager. Often with a delay, because the spreadsheet still needs to be checked.
7. Phone call. Someone answers, reads it, and calls the designer. Only now does the decision process begin.

In a continuous model, the path is shorter and does not include a human as the data carrier:

1. The logger reads the sensors in a defined cycle, in Inclify systems every 15 minutes by default, or more often (this is a logger setting; the platform accepts another step and calculates completeness against the actual cadence).
2. The hub or logger sends a frame with the readings and a timestamp to the platform over the internet.
3. The platform stores the reading, recalculates it through the project equations into an engineering value (with temperature compensation, relative to the zero reading), and compares it with the threshold.
4. The threshold WARNING or ALARM changes the channel state.
5. SMS, email, and an in-app notification reach authorized people. Someone confirms they have taken over the case, with name and time.

Worst case in this cycle: an exceedance just after a reading waits for the next one, so just under 15 min, plus transmission and recalculation. For simplicity, the whole text counts full 15 min. This is a rounding up, not down.

Clock 2: from question to answer

The “from question to answer” clock measures how much time passes from the moment someone asks, “What is happening with the asset?” to the moment they have a data-based answer. The question may come from a supervising inspector after heavy rain, a neighbor after a vibration event, the board after a media report, or a designer before the next excavation stage.

In a periodic model, the answer is: “the last reading was three weeks ago, the next one is in a week, and we can speed it up, the team can be there the day after tomorrow.” The clock shows days. In a continuous model, the answer is to open the chart: the latest point is from a quarter of an hour ago, the trend covers the last 30 days, and temperature is on the second axis as context. The clock shows minutes, exactly as long as it takes to log in and look at the dashboard.

This clock most often determines the relationship with the client and the authority: someone asked, and you answer immediately, with date and time.

Clock 3: from data to report

The “from data to report” clock measures how much work separates raw readings from a document that can be shown to someone outside the team. In a periodic model, this is manual processing: gathering logger files, harmonizing units, applying temperature compensation, creating charts, writing notes, and correcting the draft. With a dozen sensors, it is hours; with several dozen and several assets, it is working days in each cycle.

In a continuous model, the platform calculates continuously, and the report is a view of data that already exists: trend, period comparison, channel states, alarm history, CSV export. This clock does not have dramatic multipliers, but it is the one that generates the labor hours nobody remembers when signing a contract for periodic measurements. If you compare quotes, calculate it separately. It is what decides whether “cheap measurement” is really cheap.

Multiplier table - arithmetic, not a client benchmark

The figures below are division. The periodic measurement interval, expressed in minutes, is divided by 15 min. These are not results measured at any client, and they do not depend on sensor type or asset. If your schedule is different, insert your own interval. The formula is in the section “Calculate your own multiplier”.

Three notes on the table.

First, the multiplier describes how often new information appears, that is, the upper limit of how long an exceedance can remain unnoticed. It does not describe how fast someone reacts. Reaction is a separate loop, dependent on procedure and people.

Second, here is the answer to the objection that management usually raises: “Why do we need a quarter of an hour if the decision still takes two hours?” Reaction time adds to both models in the same way. Two hours after 15 min is still the same morning; two hours after a week is still a week. The multiplier measures the blind period, not the performance of your team.

Third, the table deliberately leaves out the “processing time” from the third clock. If we added it, the periodic measurement multipliers would only grow. We omit it so the comparison stays conservative and reproducible with a calculator in a phone.

That is where “100x” comes from. It is not a number from a specific deployment, but the lowest order of magnitude that comes out of this table for anything less frequent than daily measurement. If your schedule is quarterly, “100x” sounds modest. When someone shows you a larger number, ask for the divisor. Without the divisor, it is not a benchmark, just a slogan.

What shortening the loop from weeks to minutes changes - three hypothetical scenarios

The scenarios below are invented for this text. They do not describe any real asset, client, or incident. Their purpose is to show how the same physical fact leads to a different decision depending on which clock you use.

Scenario 1: an excavation in a city center

Imagine a construction manager running a deep excavation next to a historic building. A chain inclinometer is working in the diaphragm wall, and tilt sensors and crack gauges are installed on the historic building. The designer has set the warning value for wall movement as a fraction of the design value. As an illustration, let us assume the WARNING threshold is 60% and the ALARM threshold is 80% of the calculated allowable displacement.

In a weekly measurement model: on Wednesday morning, the excavator removes the layer below the level of the strut, which has not yet been installed because the steel will arrive on Friday. The wall accelerates. The next measurement is on Monday. For five days, work continues normally because nobody has reason to stop it. On Monday, it turns out the warning value was exceeded on Wednesday and the alarm value on Friday.

In a continuous model: the 7:59 reading shows an increase, still below the threshold. The 8:14 reading exceeds the threshold, the channel changes to WARNING, and an SMS reaches the construction manager and the geotechnical engineer at 8:14. At 8:30, the manager stops excavation in that section and calls the designer. The steel for the strut still arrives on Friday, but the excavation sits for those two days instead of deepening. That is the difference between “stopped at 8:14” and “stopped after the next reading”.

The point is not that monitoring “saved” the historic building. The point is that the decision was made on data from a quarter of an hour ago, not five days ago, and that there is a record of who made it and when. How much that record matters later is covered in the text about data from monitoring as evidence in a dispute.

Scenario 2: a hall under snow

Imagine a manager of a warehouse hall with a building footprint above 2 000 m². Article 62(1)(3) of the Polish Construction Law requires periodic inspection for such buildings at least twice a year, by 31 May and by 30 November, and point 4 of that provision requires an inspection of safe use each time after external factors from Article 61, such as heavy precipitation. An inspection is a point on the time axis. Snowfall, especially thaw with rain on accumulated snow, is an event between points.

In a “twice a year inspection” model: the manager checks the forecast, sends someone to the roof with a shovel or does not, and decides based on intuition. There is no number to compare with the design.

In a continuous model: vibrating wire strain sensors work on selected beams, and on the panel next to the strain chart you can see temperature and, as context, IMGW weather data from an external widget. The WARNING threshold is set as a fraction of the strain corresponding to the design load, illustrative value 70%. When strain rises at night and exceeds the threshold, an SMS reaches the manager and the structural engineer. In the morning, the manager does not guess, but knows which beams, by how much, and since when. The decision to clear snow or to take part of the hall out of use has a basis that can be shown during an inspection. We explain how this links to the obligations under Article 62 in the text about periodic inspections and continuous monitoring of halls.

Scenario 3: a bridge during reconstruction

Imagine a road manager carrying out the reconstruction of a bridge structure without closing traffic. The works include replacing bearings and lifting the span by several millimeters. The reconstruction design sets allowable height differences between supports and allowable inclinations.

In a periodic model, the surveyor measures leveling before and after each lifting operation. During the lifting itself, when the span is most sensitive, nobody has a continuous picture.

In a continuous model, tilt sensors on the span and supports read in a short cycle, and thresholds are set for the difference between adjacent supports. The works manager sees on the chart how the span responds to each lifting step. If the difference approaches the threshold, they receive a warning before the operator lifts the next millimeter. After the operation, there is a record: when, how much, and with what inclination.

In each of the three scenarios, the physical fact is the same. The difference is whether the information arrived during the event or after it, and whether you made the decision or the decision was made for you.

Where “faster” is not enough

It would be unfair to end with multipliers. Shortening the loop is necessary, not sufficient. Five things that waste even the best clock.

Threshold quality. A threshold set by eye, or too tight, produces false alarms, and false alarms teach people to ignore SMS messages. A threshold set too loose will not work when it should. Threshold values should come from the design, standards, or baseline-period statistics, and they should be tuned after the first weeks of data. We describe how to do that in the text about warning and alarm thresholds.

Procedure. An SMS at 8:14 means nothing if nobody knows who receives it, what they are supposed to do, and whom to escalate to. A procedure is a matrix: channel state -> person -> action -> deadline. Confirming takeover of a case and silencing an alarm with a time limit is not convenience, but a record that someone has accepted responsibility.

People. “What if the SMS comes at 3 a.m.?” That question comes up at every meeting, and it is valid. Someone must be authorized, available, and trained. If SMS messages go to one person who is on holiday, the first clock still shows a beautiful 15 min, while the real clock shows two weeks. The duty roster and recipient list are part of deployment, not an add-on.

NO_DATA. No data is a separate state, not zero. A sensor may fail, the hub may lose signal, or an excavator may cut the cable. If the platform does not distinguish “value within range” from “I do not know”, silence looks like calm. The NO_DATA state must have its own notification and its own procedure. Monitoring that stays silent because there is no data is worse than periodic measurement, because it gives the illusion of control.

Calibration. A sensor with an expired calibration or without correct temperature compensation measures quickly and badly. Speed does not fix drift. That is why, alongside thresholds, you need a review of the sensors themselves: what is missing, what is drifting, and what needs to be re-referenced to the zero reading.

None of these five things is a software feature. They are deployment features. The platform can help, but it will not do them for the team. For you, that means one thing: beyond “how fast”, ask the supplier what happens when nobody answers, when data are missing, and when the threshold turns out to be wrong.

What this looks like in Inclify

In Inclify, the three clocks have their counterparts in the panel. Clock one: loggers send readings every 15 minutes by default, or more often; each channel has WARNING and ALARM thresholds with hysteresis and OK / WARNING / ALARM / NO_DATA states; and SMS, email and in-app notifications reach authorised people according to their settings. A case can be confirmed, with visibility of who took it over and when, and it can be silenced, always for a defined period, never indefinitely. Missing data has its own alarm type with a configurable window, so silence from the sensor does not pretend to be calm.

Clock two: project dashboard, multi-axis charts (for example, tilt against temperature), trends and time comparisons, contextual widgets with external data (including IMGW) - the answer to “what is happening” is to open the view. Clock three: engineering reports calculated from data (calibration debt, temperature compensation, 7/30-day risk assessment, alarm tuning with threshold proposals, data SLA, meaning completeness and gaps) and CSV export. There is also a communication log with raw frames from devices (stored by default for 7 days, configurable period), which shows what arrived from the logger and when. For vibration events, a separate dynamic path with an envelope alarm is active, responding to a triggered event, not to the interval.

What the platform will not do for you, we say plainly: there is no escalation chain to additional people after a missing confirmation, no alarm for rate of change, and no push notifications in the mobile app. Who keeps the phone by the bed and what they do after the SMS remains in your procedure. Details: /platform/monitoring. If you are comparing data approaches across platforms, see the Vista Data Vision and Inclify comparison. There we compare solution classes, not speed.

Calculate your own multiplier

The multiplier is a quotient: how long you wait for information today, divided by 15 min. Calculate it in four steps.

Step 1. Write down the current measurement interval. How often does a new reading really appear today, not in the contract, but in practice? Convert to minutes.

• day = 1 440 min
• week = 10 080 min
• month (30 days) = 43 200 min
• quarter (90 days) = 129 600 min

Your interval: ________ min

Step 2. Add processing time. How much time passes from measurement on site to the moment the result is on the desk of the decision-maker? Travel, logger, spreadsheet, email. Convert to minutes. If you do not know, cautiously assume one working day (480 min) or enter zero. Then the multiplier will be conservative.

Your processing time: ________ min

Step 3. Divide by 15.

multiplier = (measurement interval + processing time) ÷ 15 min

Step 4. Compare with the table. Below 96, you measure more often than once a day, which is rare with manual measurements. Between 96 and 672, the daily-to-weekly range. Above 672, between a week and a quarter.

Illustrative example (assumed numbers, not measured). An asset measured once a week, with results reaching the contract manager after three working days of processing:

• interval: 10 080 min
• processing: 3 × 480 = 1 440 min
• multiplier: (10 080 + 1 440) ÷ 15 = 768x

The same asset measured once a quarter, with one week of processing (5 × 480 = 2 400 min): (129 600 + 2 400) ÷ 15 = 8 800x.

The result does not say whether you need continuous monitoring. It says how long your blind period is today, and whether you want to stand before an inspector, a neighbor, or the board with that period when the question comes: “Since when did you know about this?”

If your asset is designed using the observational method, this arithmetic has another meaning. PN-EN 1997-1 (Eurocode 7), section 2.7, requires the monitoring plan to reveal exceedances early enough to allow planned emergency actions, and the response time of instruments and result analysis procedures must be short enough relative to the possible evolution of the system. The measurement interval is therefore a design parameter, not a convenience for the contractor (in the second generation of Eurocode 7, EN 1997-1:2024, the observational method has different clause numbering). We write about this separately in the text on the observational method in practice, and the full topic is organized in the complete guide to structural monitoring.

Sources and further reading

• Act of 7 July 1994 - Construction Law, Articles 61 and 62 (owner and manager obligations, periodic inspections), consolidated text in ISAP: https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU19940890414 (legal status as of 22.08.2026).
• PN-EN 1997-1 Eurocode 7: Geotechnical design - Part 1: General rules, section 2.7 - observational method (PKN, sklep.pkn.pl); EN 1997-1:2024 - second generation, with different clause numbering.
• Borecka A., Stopkowicz A., Sekuła K., “Metoda obserwacyjna i monitoring geotechniczny w świetle przepisów prawa…”, Przegląd Geologiczny 2017, vol. 65, no. 10/2 - discussion of the requirements of section 2.7 in Polish.
• PN-B-02170:2016-12 - Assessment of the harmfulness of vibrations transmitted through the ground to buildings (PKN, sklep.pkn.pl) - for the dynamic path mentioned in the text.
• Other posts in the series: guide to structural monitoring, warning and alarm thresholds, periodic inspections and continuous monitoring of halls, observational method in practice, data from monitoring as evidence.

What next

Calculate your multiplier using the formula above and note three things next to it: the current interval, the processing time, and who receives the information about the exceedance today. If the result concerns you, the lowest entry point is a pilot on one asset, preferably one that has an excavation phase, winter, or an inspection ahead, because then the blind period costs the most. If the asset already has sensors and loggers, connecting it to the platform takes a few days; a full deployment with installation takes from several to a dozen or so weeks. Want to see how the same loop looks in the panel, on an asset similar to yours, with your thresholds and your SMS recipients? Arrange a short conversation: let's talk. We respond within 24 hours.