A good TARP alarm response plan combines a specific trigger condition with data quality checks, an accountable owner, an action, a confirmation deadline and a return-to-normal-work condition. A threshold in a table is not enough. If, after an alarm, the team is only then looking for the designer's phone number and wondering who can stop the works, the plan does not operate in practice.
In brief
- Every trigger condition must lead to a pre-agreed sequence: verify the data, notify the owner, carry out the action, record the decision.
- The first 15 minutes are for securing the situation and confirming facts, not diagnosing from a single chart.
- Confirming an alarm means "someone has taken ownership", not "the problem is solved".
- The return-to-work condition must be defined before the event. It should identify the authorised person, the required evidence and the scope of additional observation.
- The measurement platform delivers the alarm, history and reaction trail. It does not replace the designer, the manager or the organisational procedure.
TARP is a procedure, not a coloured table
TARP (Trigger Action Response Plan) is a plan that assigns pre-agreed actions to defined levels or trigger conditions. A trigger condition may be exceedance of a value, loss of data, an unusual site observation, or a combination of several indicators. What matters is that it is measurable and can be recognised in time for the response to still make sense.
The Global Industry Standard on Tailings Management defines TARP as a risk control management tool. There, trigger levels result from operational objectives and the risk management plan, and actions are defined in advance and depend on the size of the exceedance. Although the standard applies to tailings facilities, the logic is useful for an excavation, tunnel, embankment, bridge or hall: the measurement result must trigger a concrete organisational response.
The same assumption underpins the observational method. A JRC and CIRIA publication describes it as a continuous, managed process of design, construction control, monitoring and review, with pre-prepared modifications. So the point is not to "watch what happens". It is to prepare variants before site conditions force the team to act under pressure.
The most common failure looks harmless. The documentation contains a green-amber-red table, next to names and a generic instruction to "notify supervision". But nobody knows who should check the credibility of the reading, how long they have, whether the works may continue, what data must be collected, or who can lift the restriction. Such a document looks fine at handover. At 06:14 in the morning, it is of little use.
The lesson for you is simple: assess the TARP by whether a new shift hand can carry out the first steps without calling the document author.
Six fields that must not be missing from a single TARP row
A plan row should read from left to right like an instruction. First the condition, then verification, responsibility, action, record and return. If one cell contains five unclear sentences, split it. In practice, a good row is short, but it leads to a more detailed action card.
| Field | What to enter | Acceptance question |
|---|---|---|
| Trigger condition | Measurable condition and data source | Do two people recognise it in the same way? |
| Data check | Quick reliability checks | What excludes a time, transmission or sensor error? |
| Owner | One role on shift | Who takes the alarm and confirms receipt? |
| Action | Action appropriate to the level | What exactly may, must or should be stopped? |
| Evidence | Required record | Where are the time, person, data and decision retained? |
| Return | Closure criterion | Who restores the normal mode, and on what basis? |
The trigger condition should identify the parameter, level and context. "Large displacement" is an opinion. "Exceedance of the agreed alarm level for the control point during excavation of section B" is a recognisable condition. For a no-data alarm, you must define the acceptable window based on the real measurement cadence. More on this distinction is described in the article no data is not a safe state.
Quality control must not turn into a one-hour investigation. At first, you check the time stamp, freshness of adjacent channels, continuity of the series, temperature, communication status and consistency between the raw reading and the calculated value. If these checks do not settle the matter, treat the alarm as potentially real. Do not reset the reference or move the threshold just to make the colour return to green.
The owner should be the on-duty role, not a specific name embedded in a PDF. Names change with the rota, while responsibility must survive leave and contractor change. The contact list can be a separate, controlled appendix.
The first 15 minutes: 0-5-15 checklist
The rhythm below is an organisational pattern, not a universal standard requirement. The timings must be adapted to the speed of the phenomenon, the structure and the works. For a sudden water inflow, even five minutes may be too long. For slow ground consolidation, the same period mainly helps organise information.
Minute 0: the alarm becomes a task
The on-duty person confirms receipt of the alert and checks which asset, point and level the alarm concerns. They read the applicable TARP card. If the trigger condition requires immediate stoppage of a specific activity, they do not wait for a full diagnosis. They issue the instruction in line with the procedure and record the time.
Up to 5 minutes: rapid reliability check
The on-duty person checks:
- whether the reading has a current and unambiguous time stamp;
- whether there are consecutive samples before and after the exceedance;
- whether the adjacent channel and temperature data are fresh;
- whether a no-data alarm occurred at the same time;
- whether the raw value and the equation result refer to the same moment;
- whether the construction log contains an event that could explain the change.
The check is meant to expose an obvious data error. It is not meant to invalidate an alarm when the result is inconvenient.
Up to 15 minutes: responsibility and the next decision
The alarm owner passes a set of facts to the person named in the TARP: value, threshold, time, rate of change, data quality, status of adjacent points and actions already taken. They agree the next review time. If a site check is needed, they assign it to a specific role and specify what that person should look for. Every decision goes into the event log.
0-5-15 acceptance checklist
- [ ] The event has one owner.
- [ ] Receipt confirmation has a time and a person.
- [ ] Rapid data quality checks were performed.
- [ ] The action assigned to the level was applied, without improvising the threshold.
- [ ] The notified people and issued instructions were recorded.
- [ ] The next review time was agreed.
- [ ] Data from before and after the event were retained.
For you, this checklist is a practical test. If any field cannot be completed before the alarm happens, the process must be improved rather than relying on the on-duty person's reflexes.
Illustrative example: excavation wall displacement
Assume, for illustration, that the designer has set two response levels for the displacement of a deep excavation wall. These are not values recommended for other assets. The numbers in a real project result from calculations, ground conditions, the construction stage and the adopted excavation method.
At 06:14, one point exceeds the warning level. The reading is 1.2 mm above the threshold. The previous three samples were rising, the data are fresh, and the adjacent point shows a change in the same direction. Temperature is stable. The construction log notes night-time deepening of the section at that wall.
| TARP row element | Illustrative example |
|---|---|
| Trigger condition | Exceedance of the warning level in the active section |
| Check | Time, three previous samples, adjacent point, temperature, communication |
| Owner | On-duty monitoring engineer |
| Action | Stop deepening the section, notify the manager and designer |
| Evidence | Chart export, raw reading, confirmation, works log entry |
| Return | Written decision by an authorised role after assessing the data and site |
This is not yet a diagnosis. The correlation between two points and the construction stage increases the credibility of the event, but the cause still needs assessment. The team may order a control measurement, inspection, check of support and comparison with the expected behaviour. What matters is that work in the most sensitive section does not continue for hours simply because nobody received the message.
The Crossrail case of diaphragm wall monitoring shows the same logic in a real project: trigger levels were linked to communication, data review, contingency plans and a change in the way the works were carried out. The publication does not provide universal thresholds or timings for your asset. It does, however, show clearly that measurement was part of construction control, not decoration for a report.
Confirmation, silencing and closure mean three different things
These three operations are often mixed together, and then the event history becomes unreadable. Confirmation means that the nominated person has taken over the alarm. It does not change the measurement value and does not restore an OK state. Silencing stops further notifications for a defined period, for example during a planned test or maintenance work. Closure means that the condition set out in the TARP has been met and the response mode can end.
| Operation | What it confirms | What it does not confirm |
|---|---|---|
| Confirmation | A person has taken ownership of the event | That the cause is known or removed |
| Silencing | Notifications are temporarily limited | That the measurement has returned to normal |
| Return to work | An authorised role has accepted the evidence | That the risk has disappeared forever |
The most risky option is silencing without a deadline and without an owner. After a few days, nobody remembers why the alarm stopped arriving. That is why every silencing period should have an end, a reason and a person responsible for reassessment. We describe mechanisms for limiting repeated notifications separately in the text on hysteresis and reducing repeat notifications.
The return condition should be just as precise as the trigger condition. "When things settle down" does not resolve anything. A better entry identifies the required control measurement, the period of stable observations, the inspection, the designer's consultation and the person issuing the decision. The stabilisation time must come from the nature of the phenomenon, not from the convenience of the schedule.
What it looks like in Inclify
Inclify handles the technical part of the alarm process. For a channel, you can set WARNING and ALARM levels with hysteresis, and a separate alarm type detects no new data. Authorised users receive email, SMS and in-app notifications according to their preferences.
After taking over an event, the user can confirm the alarm. The platform records the person and the time. An alarm can also be silenced only for a defined period, and the history of state transitions remains visible. Alarm configuration changes are logged in the audit trail with before-and-after values. This lets you verify whether the threshold was moved after the event without leaving a trace.
The platform does not create an organisational escalation chain to additional people after no confirmation. It also does not define, on behalf of the designer, site actions or the condition for resuming the works. Those elements must be included in the TARP and in the team's responsibilities. Inclify provides the current state, notification and response record, but a person still makes the decision. More on level configuration can be found in the guide how to set alarm thresholds.
Before launching the TARP, it is worth running a test scenario for each notification channel and confirming who will see the event outside standard working hours.
TARP limitations that must be stated before implementation
TARP will not fix a poorly designed monitoring programme. If the sensor is in the wrong place, the threshold does not correspond to the hazard mechanism, or the reading interval is slower than the development of the phenomenon, the procedure will react to late information. First define the engineering question, then the measurement, and only then the response.
The plan also does not remove measurement uncertainty. One unusual point may result from a real change, a fault, temperature, a reference error or timing. The procedure should require data checks and site observation, but it cannot assume that every doubt can be resolved remotely.
The next limitation is organisational. TARP works only when the person on shift knows their role, has access to the data and can issue the planned instruction. A document without drills ages quickly. After a change in construction stage, support geometry, contractor, contact details or design assumptions, it should be reviewed.
Finally, there is responsibility. The platform can show an alarm and record confirmation, but it does not give the user construction authority and does not assess structural safety. In an unclear situation, the safe option should be maintained until the appropriate person has assessed it.
TARP implementation checklist for a project workshop
Before the monitored phase starts, bring together the designer, manager, monitoring engineer, supervision and the contractor's representative responsible for the works. Go through each trigger aloud. A good workshop ends with answers, not another version of a coloured table.
- [ ] Each trigger condition has a unit, source, level and applicable stage.
- [ ] The response to no data and to an unreliable time has been defined.
- [ ] For each shift, one owner role and one deputy are assigned.
- [ ] The on-duty person has access to the chart, raw data and contacts.
- [ ] The first actions are feasible at night and at weekends.
- [ ] It is clear who may restrict or stop specific works.
- [ ] Alarm confirmation is distinguished from event closure.
- [ ] Silencing has a deadline, a reason and an owner for reassessment.
- [ ] The return condition identifies the evidence and the approving person.
- [ ] The team has run a drill with an alarm, no data and an incorrect time.
- [ ] After the drill, the procedure and contact list were updated.
- [ ] A TARP review has been scheduled after a change in the work stage or configuration.
FAQ
Is TARP required for every monitoring system?
There is no single universal legal rule for all assets and sectors. However, TARP is a practical way to link monitoring with risk control, especially in the observational method and in works with rapidly changing conditions. The scope of the procedure should follow from the design, contract, risk assessment and the requirements applicable to the specific asset.
Who should set the thresholds in the TARP?
The thresholds should be defined by a competent designer or by the team responsible for safety and structural behaviour. The platform provider can help implement the values and check the alarm logic, but should not invent the technical limits alone. Every threshold needs a source, a unit, an applicable stage and the associated response.
Can confirming an alarm automatically close the event?
It should not. Confirmation only means that a specific person has seen the alarm and taken responsibility for the next process step. Closure requires the conditions in the TARP to be met, for example a site inspection, an independent measurement, a period of stable data and the designer's decision. Combining these operations hides unresolved events.
What should be done when an alarm is accompanied by no data from other channels?
You need to increase the uncertainty of the assessment, not treat missing measurements as proof of calm conditions. Check communication, the clock, the last valid samples and data from independent points. If reliability cannot be confirmed quickly, apply the safe action set out in the TARP until data are restored or the site has been checked.
How often should the response plan be rehearsed?
The frequency depends on the risk and the pace of change on the project. It is worth running a drill before launching a critical phase, after a change in people or contacts, and after any significant change in the structure, support or monitoring configuration. Every real alarm and every mistake is also a reason for a short procedure review. The first end-to-end trial is worth including in the FAT and SAT of a vibration monitoring system, so that you test not only the message, but also the team's response.
Can TARP replace the observational method?
No. TARP is one of the execution tools. The observational method covers design assumptions, the expected behaviour range, the monitoring plan, prepared variants and the ability to implement changes safely. We describe it in more detail in the article on the observational method under Eurocode 7. TARP organises the moment when these pre-prepared actions are triggered and assigns operational responsibility.
Sources and further reading
- Health and Safety Executive, Safety of New Austrian Tunnelling Method Tunnels.
- Joint Research Centre / CIRIA, The Observational Method in ground engineering: principles and applications.
- Crossrail Learning Legacy, The Use of Shape Accel Arrays for Measuring Retaining Wall Deflection.
- Global Tailings Review, Global Industry Standard on Tailings Management.
- Joint Research Centre, Guidelines for reliability-based verification of geotechnical limit states, część o metodzie obserwacyjnej.
What next
Take one existing TARP row and try to run it according to the 0-5-15 checklist. If you cannot identify the owner, evidence or return condition, the gap is already visible. Arrange a conversation with the Inclify team if you want to translate the alarm procedure into a working flow of notifications, confirmations and history for one pilot asset.