Choose a site for a structural monitoring pilot where there is a real question to answer ("can we keep excavating?", "what is the roof doing under snow?"), access for installation or existing sensors, a named process owner, and time to see a trend over at least one construction phase or one season. Not the largest asset in the portfolio, but the one where a decision based on data can be shown in 90 days.
In brief
- A pilot has two goals: proof for management and learning for the organisation (thresholds, procedure, people). Pretty charts alone are not enough.
- Six site-selection criteria scored from 1 to 5: real risk, access and installation, existing sensors, process owner, duration, and visibility to decision-makers. The weakest link decides, not the total.
- 90 days is enough if you have a zero reading, preliminary thresholds and a written response plan before you start. The plan below is a recommended template, not a rule.
- Measure success by data completeness, time from event to SMS, number of false alarms after tuning, the procedural trail and whether the report reached management.
- The most common pitfalls are too many sensors, no zero reading, no owner, and a pilot with no end criteria.
Why run a pilot at all
A structural monitoring pilot is a time- and scope-limited deployment of a measurement system on one asset, designed to answer whether continuous monitoring changes decisions in the organisation and whether it is worth scaling. The pilot does not test the technology; it tests the organisation.
If you are about to start monitoring in a company, you are probably worried about one thing: that you will choose the wrong site, nothing will show up for three months, and the organisation will be left with the sentence "we tried, it does not work" - and the topic will return only after a failure. That concern can be controlled. You need three things written down before the decision: site-selection criteria, success metrics and an end date. The rest of this text is built around them.
A pilot has two goals. The first is proof for management. Management does not buy sensors; it buys the sentence "we knew earlier, and we have evidence for it." After the pilot, you must be able to show: here was the event, at this time the SMS arrived, this person confirmed the case, this is how we responded. If nothing happened over 90 days, that is also fine: you show that the asset worked within the assumed range, and daily thermal "breathing" did not trigger alarms.
The second goal is learning for the organisation. The first thresholds are always too tight or too loose. The first procedure has gaps: the SMS comes on Saturday at 3 a.m., and nobody knows who should go. The first report is too long for management and too short for an engineer. In a pilot, those mistakes cost little; in a roll-out across ten assets at once, they cost a lot and damage confidence in the whole idea.
If you are still building the budget case, start with the post on how to justify structural monitoring to management - the pilot is its natural next step. The broader context is covered in the complete guide to structural monitoring (SHM). For you, this means one thing: the pilot wins or loses at the site-selection stage and in the written rules, not at the sensor-installation stage.
Site-selection criteria
The most common mistake is to choose the largest asset, the most prestigious one, or the one that "happens to have budget". The largest asset has the most stakeholders and the longest approval path, so the pilot can get stuck before the first sensor sends a reading. Instead of prestige, use six criteria on a 1-5 scale. Assess two or three assets at once, on one sheet.
| Criterion | Why it matters | How to score it (1-5) |
|---|---|---|
| Real risk / question to resolve | Monitoring should improve a specific decision, not create data. Without a question, there is no success criterion. | 1 = "we want to see how it works"; 3 = known issue, but no decision date; 5 = a specific decision with a date (excavation phase, winter season, refurbishment stage) |
| Access and installation feasibility | Installation under traffic, at height or with a lane closure shifts the start by weeks and increases cost. | 1 = closures, permits, a lift required; 3 = access in night windows; 5 = ground-level access or existing platforms, power nearby |
| Existing sensors / loggers | If the asset already has sensors and a logger with a data output, connecting it to the platform takes days instead of weeks. | 1 = nothing installed; 3 = sensors are present, but the logger has no data output or no documentation; 5 = logger/hub capable of sending HTTP/JSON data and access to reading history |
| Committed process owner | Someone must receive SMS messages, confirm alarms, tune thresholds and take the report to management. Without that person, the pilot dies after a week. | 1 = "the technical department will handle it"; 3 = there is a person, but no time allocated in the schedule; 5 = a named owner with the authority to stop works or close the zone |
| Duration | You only see a trend when you can distinguish it from daily and seasonal temperature variability. The pilot must cover at least one change in conditions. | 1 = works end before the first data arrive; 3 = 90 days with no clear phase change or season change; 5 = 90 days covering a work phase or a seasonal transition |
| Visibility to decision-makers | A pilot that management does not know about will not convince management. The asset should matter to someone who signs the budget. | 1 = peripheral asset, no sponsor; 3 = important for one department; 5 = an asset that management asks about in meetings (claims, contractual penalty, reputation) |
How to read the result
Add the points up, with a maximum of 30, but first look at the lowest score - the weakest link decides the outcome. Use a simple rule: any score of 1 or 2 excludes the asset from the first pilot, even if it scores highly in the other criteria; all scores of at least 3 and a total of 22 or more make it a good candidate; a total below 22 without any 1s or 2s makes it a reserve candidate, and one criterion needs to be strengthened, for example by adding a management sponsor.
Two dilemmas come up most often. The first is fast start (existing sensors, good access) versus real risk. Choose risk - a pilot on an asset where the data do not decide anything will start quickly and prove nothing. The second is owner versus visibility. Decide in favour of the owner; a management sponsor can be added at the reporting stage, but a missing owner cannot be added halfway through. Keep the written scoring - it is the first appendix to the pilot contract and a ready answer when someone asks why this asset.
Three typical good pilot assets
These are scenarios, not deployment descriptions; run each through the six criteria in the table.
An urban excavation with neighbours
Imagine a site manager with a three-level excavation, a diaphragm wall and a masonry building a few metres from the edge. The question is clear: is the neighbouring building responding to the next excavation phases, and can we keep excavating. The work phase gives natural time boundaries, for example excavation down to the brace level and the first anchoring. Sensors: a chain inclinometer (in-place) in the diaphragm wall, tilt sensors on the masonry building, crack gauges on existing cracks, and possibly a piezometer. The process owner is the site manager, who has the authority to stop the excavator; visibility is provided by the contract director, because a neighbour claim is their risk. What to document and how it protects you in a dispute is covered in the text on monitoring neighbouring buildings during construction.
Weak points: installation on someone else's building requires the owner's consent, and the zero reading must be taken before excavation starts, not during it.
A hall with a large roof
A warehouse, production or sports hall manager with a large-span roof. The question is: what happens to the trusses between periodic inspections, especially under snow. Strain vibrating wire sensors on selected trusses, a deflection sensor, and temperature as background. Access is usually good (a lift, technical platforms), and power is available on site. Plan the pilot so that it covers winter - without snow you will only see thermal breathing. The owner is the facility manager; visibility is provided by the person responsible for the periodic inspections under Article 62(1) of the Building Law - for large halls (buildings with a building footprint area over 2 000 m²), there are at least two inspections per year, by 31 May and by 30 November, and between them the roof has no measurement-based supervision.
Weak points: a snow-free winter will not show an event. It will, however, show the trusses' response to temperature and whether the thresholds make sense - and that is already a result.
A bridge or flyover under refurbishment
A road authority with a structure under reconstruction: bearing replacement, suspension, launching, proof loading. The question is whether the structure behaves according to the design at successive work stages. Sensors: support tilts, girder strains, bearing and expansion joint displacements, accelerometers to observe natural frequencies. The work phase gives a clear "before / after" and a ready schedule. The owner is the supervising engineer; visibility comes from the road authority, for whom traffic stoppage is a cost measured in hours.
Weak points: installation is often under traffic or with lane closures, so the access score drops. It pays to coordinate installation with the contractor's works, because they already have scaffolding and closures in place. If none of the three scenarios fits your portfolio, that is fine: the criteria table works for any asset, and the scenarios only show what a complete case looks like - question, phase, owner, sponsor.
90-day pilot plan - recommended template
The plan below is a template we recommend as a starting point. Adjust the phase lengths to the works schedule and the season, but do not remove any phase - each one supports one of the success metrics.
A zero reading is the first full set of measurements taken after installation and sensor stabilisation, but before the event whose impact you want to measure; all later values are calculated as a difference from it.
We count 90 days from the first reading, not from contract signature. If the asset already has sensors and a logger, the connection takes a few days and the pilot starts almost immediately. With installation from scratch, add several to a dozen weeks of preparation - and make sure the 90-day data window falls within a work phase or season.
| Phase | What we do | What we expect | Who |
|---|---|---|---|
| 0. Preparation (before day 1) | Write down the pilot question, end criteria, a short list of measurement points, preliminary thresholds from the design or standard, and the response plan (who gets the SMS, who confirms, who decides). Install or connect the existing loggers. | A one-page "pilot contract" signed by the owner and sponsor. A zero reading. | Process owner, SHM engineer / designer, supplier |
| 1. Days 1-30: baseline period | The logger sends readings (every 15 minutes by default, or more often), thresholds are not changed, and we observe daily thermal breathing. First check: whether every sensor reports. | Data completeness; chart of the measured quantity against temperature; list of channels with NO_DATA and their causes. | SHM engineer, supplier |
| 2. Days 31-45: tuning | Using the baseline data, verify the WARNING / ALARM thresholds and hysteresis, set the NO_DATA alarm window, and define the SMS and e-mail recipient list. Procedure test: simulated exceedance, SMS, confirmation, note. | Thresholds that do not trigger on temperature but do trigger on an out-of-range change. Measured human response time. | SHM engineer, process owner |
| 3. Days 46-80: event observation | The pilot runs in the target work phase or season. Every alarm: confirmation, decision, note. Monthly report. | At least one real event or condition transition; documented response. | Process owner, site team |
| 4. Days 81-90: assessment and decision | Final report: success metrics against the phase 0 criteria, recommendation (scale / improve / end), deployment model for additional assets. Management presentation. | A management decision, not "another meeting". | Process owner, management sponsor |
Why the baseline period cannot be skipped
Thresholds set before seeing the asset's natural variability are guesswork. The design provides limit values, but it does not say how much a girder, wall or support "breathes" over a day, or what noise a specific sensor has in a specific location. Thirty days of readings every 15 minutes by default, or more often, are enough to see the daily cycle and several weather changes, and on that basis set meaningful thresholds. If the asset already has a logger with stored history, part of the baseline period can be built backwards - more on that below. How to derive thresholds and avoid false alarms is covered in the text on setting warning and alarm thresholds.
Pilot success metrics
Define them in phase 0 and write them into the pilot contract. Five metrics that stand up to management and the engineer:
- Data completeness (data SLA). What share of expected readings reached the platform, broken down by channel. NO_DATA gaps are the first thing an auditor or expert witness will ask about. More important than the count is whether you can explain each gap (power, transmission, cable, service).
- Time from event to SMS. Measured in the phase 2 test and during real exceedances. With readings every 15 minutes by default, or more often, an event waits for the next frame for at most a quarter of an hour - that limit comes from the logger setting; in the pilot you verify that nothing extends it: transmission, wrong recipient list, unconfirmed phone number, muted device.
- Number of false alarms after tuning. Before tuning there will be many, and that is normal. After phase 2, every alarm should be a real event or have an identified cause to remove. A false alarm is one that nobody can assign to any cause.
- Whether the procedure worked. Whether someone confirmed each alarm, after how long, whether a decision was made, and whether it was recorded. The trail matters, not the impression. If the same person always confirmed it at any time of day, that is also a result: the procedure depends on one person.
- Whether the report reached management. And whether management read it. A pilot whose results get stuck in the maintenance department does not exist from the perspective of scaling.
For each metric, write into the contract the value you will treat as success. Illustratively: at least 95% completeness on each channel, every alarm confirmed within an hour on a working day, zero alarms without an assigned cause after phase 2. The numbers are yours - the important thing is that they exist before the start, not after it. What not to measure: the number of sensors, charts and logins - these are activity measures, not value measures. Five numbers on one page are your entire defence against the question "so what did it deliver?"
Pitfalls
Too many sensors
The temptation is: "since we are installing anyway, let us add a few more." Each measurement point is a cable, calibration, a channel to tune and a potential source of NO_DATA. In a pilot, what matters is the answer to the question from phase 0, not asset coverage; remove any point that does not answer it. Fewer, better chosen points are also the simplest way to control cost - the cost components are covered in the text on what structural monitoring costs are made of.
No zero reading
Installing during excavation or after the first snowfall means you do not know how much displacement existed "before"; every later value becomes open to challenge, technically and in a dispute. If you do not have time before the event, move the pilot to the next phase or change the asset.
No owner
The alarm lands in a mailbox nobody reads, and after two weeks the system is "the thing that rings". The owner must have a name, a phone number, time in the calendar and the authority to decide. If there is no such person, the first conclusion from the pilot is ready before it has even started.
A pilot with no end criteria
"We will see how it goes" ends with a pilot that runs for a year and nobody knows whether it succeeded. Write the end criteria before the start: after 90 days and the report, management decides - scale, improve or stop. A pilot without an end date is a trial period, not a decision tool. All four pitfalls are disarmed by the same thing: a one-page pilot contract from phase 0, signed before anyone touches a sensor.
What this looks like at Inclify
A pilot on one asset is one of three cooperation models with Inclify, alongside turnkey monitoring and the platform alone for existing sensors. If the asset already has a logger or hub capable of sending data via HTTP/JSON, the connection takes a few days; the reading history from the existing system can be sent through the same channel with the original timestamps, so the baseline period does not have to start from zero. With installation from scratch, count several to a dozen weeks.
The most useful pilot features are: a data SLA report with completeness, gaps and freshness per channel (metric no. 1); a no-data alarm with a configurable window; OK / WARNING / ALARM / NO_DATA states with SMS and e-mail notifications, acknowledgement, case ownership and silencing for a defined period at all times (metric no. 4); threshold proposals calculated from measurement history together with the expected number of events per week, which you apply in one action after the baseline period; a communication log with raw frames from devices (what arrived, when, and from which device); a tilt chart against temperature on two axes; and a project dashboard that management can understand without explanation. What the platform does not have: trend alarms - slow change is read from the chart and from the 7/30-day risk assessment report - or PDF reports; the management report is assembled from CSV export and saved chart images. How to prepare existing equipment is covered in the guide to connecting loggers to the online platform. Platform description: /platform.
FAQ
How long should a structural monitoring pilot last?
A sensible minimum is 90 days counted from the first reading: 30 days of baseline period, two weeks for threshold and procedure tuning, and the rest for observation in the target work phase or season. A shorter pilot will not let you distinguish a trend from daily thermal breathing; a longer one without end criteria turns into an open-ended trial period. Treat 90 days as a template and adapt it to the works schedule.
Does a pilot make sense if we already have sensors and a logger?
Yes, and it is usually fastest then. Connecting an existing logger or hub to the online platform takes a few days, so the pilot tests what is truly new: thresholds, alarms, the response procedure and the report for management. The condition is that the logger must have a data output. If it also has stored reading history, you can send it with the original timestamps and partly build the baseline period backwards.
How many sensors are enough for a pilot?
As many as answer the question from phase 0, and not one more. In an excavation with neighbours, that is usually several tilt sensors, crack gauges and an inclinometer in the wall; in a hall, strain vibrating wire sensors on a few trusses plus temperature. Each extra point is a channel to tune, a source of data gaps and a cost that proves nothing.
What if nothing happens during the pilot?
That is also a result. You show management that the asset worked within the assumed range, the system did not generate false alarms after tuning, the data were complete, and the procedure passed a simulated test. The absence of an event does not invalidate the pilot - what invalidates it is the absence of criteria against which a calm quarter can be assessed. That is why the criteria are written before the start.
Who should own the pilot - technical department or business?
The person who has the authority to make a decision based on an alarm: site manager, facility manager, supervising engineer. The technical department or supplier is responsible for sensors, data and thresholds, but will not carry the results to management. A management sponsor is needed from phase 0, because without one the final report has no recipient.
What if the pilot performs badly - does that close the topic in the company?
No, if the criteria were written before the start. Then the final report says what failed: asset selection, procedure, thresholds or hardware - and those are things to improve, not proof that monitoring does not work. The topic only "burns" if there is a pilot with no criteria, after which only an impression remains. That is why the asset scoring and the pilot contract are created before the first sensor is installed.
Sources and further reading
- Act of 7 July 1994 - Building Law, consolidated text in ISAP: https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU19940890414 (Articles 61 and 62 - owner and manager obligations, periodic inspections).
- Regulation of the Minister of Transport, Construction and Maritime Economy of 25 April 2012 on determining geotechnical conditions for the foundation of construction objects (Dz.U. 2012 item 463): https://isap.sejm.gov.pl/isap.nsf/DocDetails.xsp?id=WDU20120000463
- PN-EN 1997-1 Eurocode 7: Geotechnical design - Part 1: General rules (observational method) - PKN, sklep.pkn.pl.
- Other posts in the series: structural monitoring guide, how to justify monitoring to management, warning and alarm thresholds, connecting existing loggers, what structural monitoring costs are made of.
This text does not constitute legal advice; before making a decision, consult the current wording of the regulations with a lawyer or a person with building qualifications.
What next
You do not need to have selected a site to start. List two or three candidates and, for each one, write down: what question it is meant to answer, what is already on it (sensors, logger, access) and who would own the process. Send us that list - we will reply within 24 hours, and in the call we will go through the six criteria together, identify the weak links and propose a 90-day pilot scope for the asset with the best chance of proving something. If a hall is involved (the pilot must cover winter) or an excavation that is just about to start, timing matters: the zero reading must be taken before the first snowfall and before the first work phase, and installation or connection adds several days to a dozen weeks on top of that.