Reliability engineering for mines and heavy plant: a 24-month improvement program
Reliability improves in a fixed order: a schedule that holds, work planned a week ahead, preventive maintenance done as written, records that say what failed and at what age, and only then analysis. Pardus runs that program with your team, on site where you want an engineer there, from a 90-day diagnosis to Weibull, spares and replacement models once your data can carry them.
- The program
- Four phases over 24 months, each ending on a test
- By day 90
- Loss found, cause named, tactic changed, case costed
- To start
- Twelve months of downtime, the asset register and your PM schedules
- Handed over
- Changed plans in your CMMS, written frameworks, roles and processes, audit and monitoring routines, the KPI pack and Excel models, unlocked
Start with the schedule, not the statistics
The order is the same at every site. Get a schedule. Plan the work a week ahead. Do the preventive maintenance already committed and record what was found. Fix the records so that a work order says what failed, on which functional location, for how long and at what age. Only then start analysis.
In Pardus’s experience a scheduled maintenance program that is competently executed delivers the largest single slice of the improvement available to a site, of the order of the first half of it. Optimising that program with your own failure data, and adding condition monitoring on the modes worth watching, takes a site to perhaps eighty or ninety per cent of what is achievable. These are judgements, not measurements. The last part is where live monitoring, integrated data platforms and machine learning sit, and it costs many times more per point of availability than the two steps before it.
This is why Weibull analysis sits in the third phase of the program and not the first. It needs component change-out ages that most sites have not been recording, and a planning system that can carry out what the analysis concludes. At a site running 60% schedule compliance it produces optimal intervals that nobody performs on time. The first analysis in this program uses data every site already holds: a year of downtime, drawn as a Pareto chart and a jack-knife diagram.
Each piece of work ends in your system, not in a document. A study that ends in a report ends there. A study that ends in loaded task lists, a signed change record and a named owner for each action survives the engagement.
Six checks that show where your site stands
- Can you produce last month’s schedule compliance and planned share of labour hours, with the definition behind each?
- Pull ten closed preventive maintenance work orders at random. Do they carry meter readings, findings and any follow-up work?
- Do your records show the age at which each component was changed, or only that the machine was repaired?
- Can you name the ten losses that cost the most downtime last year, and the person who owns each one?
- Name three decisions in the last year that analysis changed, and the dollars attached to each.
- Ask a maintainer why the biggest service interval is what it is. Is the answer in their own words, and roughly right?
The maturity ladder: most of the value is on the lower rungs
Most of the achievable improvement comes from the first two steps up. Test a claim with an artefact, never with a question: ask to see the last three root cause analysis reports and what changed as a result, not whether the site does root cause analysis. Sites usually place themselves one rung above what the evidence supports, and the gap is almost always in execution rather than intent. If the first check above cannot be answered, the site is on the first rung and the next move is the schedule.
| Rung | What the site looks like | The test | The next move |
|---|---|---|---|
| Reactive | Crews chase breakdowns. Planners expedite parts for jobs in progress. PM is done when there is time. | Nobody can produce the planned share of labour hours or last month’s schedule compliance. | A protected weekly schedule, and PM treated as non-negotiable. |
| Basics | Work is planned a week ahead and the schedule is mostly met. PM compliance is above 90% within a window. Work orders close with findings. | Ten closed PM work orders show readings, findings and follow-up work. Component change-out ages exist. | Criticality and data discipline, so that effort is aimed and analysis becomes possible. |
| Optimised PM | Tactics come from failure modes and site evidence rather than the OEM booklet. Intervals are set from life data and cost. Repeat failures get a root cause analysis. | The analysis behind one interval change in the last year can be shown. | Condition monitoring on the modes with a P–F interval worth having. |
| Predictive | Condition monitoring routes run with limits and trends. Alarms are acted on inside the P–F interval. | The share of major component changes last quarter that were called by condition rather than by a failure. | Close the loop: indicators feeding a scheduled review of the plan. |
| Optimising | The plan is reviewed on a cycle against its own results. Life-cycle and capital decisions are modelled. Reliability growth is tracked. | Three decisions in the last year that analysis changed, with the money attached. | Keep going, and defend the basics, which decay the moment attention moves. |
A rung cannot be skipped, although pieces of a higher one can be bought. A predictive monitoring system on a reactive site produces alarms that nobody can act on. Statutory and safety-critical work is the floor beneath the ladder, not a rung on it: an intolerable risk, such as a failed protective device or a missing statutory inspection, is acted on now, whatever rung the site is on.
Four phases, each with an exit test
The dates are a template. A small plant moves faster and a five-pit mining complex slower, so each phase ends on its exit test and not on a date. The analysis-heavy work sits in Phase 3 because it needs component ages the site usually does not yet have, and a planning system able to carry out what it concludes.
Phase 1, diagnose: months 0 to 3
Starts with a named sponsor, a reliability engineer with protected time, and access to the CMMS, dispatch and cost data. The work is criticality, loss analysis, a data health check, a first work management audit with a baseline sample of crew time, baseline indicators, two or three root cause analyses and the first tactic changes. It ends when the loop has been proven once: a changed plan in the CMMS, a costed case approved, a live baseline pack and a 12-month roadmap placed on the ladder.
Phase 2, stabilise: months 3 to 9
Make the ordinary week trustworthy. Protect the planner and lock the weekly schedule, do the PM as written, enforce the data standard at close-out, start the monthly audit cycle and the weekly monitoring report, run the bad-actor loop on the top-ten losses, and start condition monitoring on the critical classes. It ends after a full quarter, not one good month, at schedule compliance of 80% or more, PM compliance of 90% or more in the window, planned share of 70% or more, backlog of two to four crew-weeks and close-out data quality of 90% or more.
Phase 3, optimise: months 9 to 18
Analysis on your own evidence, in a fixed order for each critical asset class: life data first, then tactics, then spares, then the capital view. Weibull on the top rotables with suspensions included, intervals re-derived against the cost ratio and grouped into service blocks, PM optimisation of the big services with the crews in the room, spares re-sized, alarm limits set from your baselines, reliability growth tracked per fleet, and life-cycle and replacement models for the fleet plan. It ends with tactics on the critical classes evidence-based and signed, growth slopes improving on the target fleets, and a benefits register banked and audited.
Phase 4, sustain: month 18 on
Management of change on every tactic, an annual strategy review, criticality refreshed, the indicator set pruned and re-targeted, and named succession for the key roles. The test is succession. If the reliability engineer resigns and the reviews still happen the following quarter, the site has a reliability system. If they stop, it had a reliability engineer.
The first ninety days
1
Month one: diagnose
Secure the mandate in week one: a named sponsor, the engineer’s time protected in writing, and access to the CMMS, dispatch and cost systems. Run or refresh criticality. Draw a year of downtime as a Pareto chart and a jack-knife diagram. Run the data health check and a first audit of work order and planning quality, with a sample of crew time. Publish the baseline pack in week four, even while the numbers are poor.
2
Month two: analyse
Take two or three top losses to root cause analysis, with the people who fix the machines in the room. Fit the top rotables, suspensions included, where change-out ages exist; where they do not, that goes on the data gap list and the recording starts that week. Review the spares for those items. Start PM optimisation on the biggest service block.
3
Month three: deliver
Load the changed tactics into the CMMS through management of change, with a signature. A changed plan that exists only in a report is a failure of the phase. Write the business case in dollars and open the benefits register, so that every later claim can be audited.
4
Day 90: the review
Show the sponsor the loop once, end to end: loss found, cause named, tactic changed, case costed. Publish the 24-month roadmap with the next quarter in detail. Nobody funds year two of a program on a promise of results in year three.
| By day | The work | What you can hold |
|---|---|---|
| 30 | Criticality. Downtime Pareto and jack-knife diagram. Data health check. Work management audit. Baseline indicators. | A ranked asset list, the top-ten loss profile, a data gap list, first audit scores for work orders and planning, and baseline availability, MTBF and PM compliance. |
| 60 | Two or three root cause analyses. Weibull on the top rotables where ages exist. Spares review. PM optimisation started on the biggest service. | Root cause actions with named owners and dates, β and η with the tactic each implies, and a spares recommendation in dollars. |
| 90 | Revised tactics loaded. Business case. Reporting running. Roadmap. | A changed plan in the CMMS, approved; a costed case; a live indicator pack; and the 24-month roadmap on the ladder, with the next quarter in detail. |
The worked example in the book is an underground loader fleet: 11 machines that lost 19,126 hours to 4,054 unplanned interventions in a year and ran at 71% availability. The jack-knife diagram ranked the motor category first, where the Pareto chart ranked it second on downtime alone.
What gets measured, and when
A program measured on everything at once is measured on nothing, so the indicators are added in layers. Each phase is governed by the layer that matches its work.
From day one: the baseline
Availability and utilisation by fleet, MTBF and MTTR with their event counts, PM and schedule compliance, planned share of labour hours, backlog in crew-weeks, the top-ten loss list with jack-knife positions, and cost per tonne and per operating hour. It is published while the numbers are still poor, because a starting point that was not recorded cannot be improved on.
Phase 2: the leading set
Schedule compliance, PM compliance in the window, planned share, backlog and close-out data quality, reported weekly, each with one named owner. Added to these is the share of root cause actions closed on time, which is the earliest leading indicator a program has.
Phase 3: the analysis layer
The share of critical-class tactics that are evidence-based and signed. Condition monitoring coverage of the highest-criticality classes, and the share of component changes called by condition rather than by failure. The Crow–AMSAA slope for each target fleet, which replaces month-to-month MTBF as the honest trend.
Phase 4, and throughout: one page and a benefits register
By Phase 4 the pack is back to one page: the lagging set plus the few leading indicators that earned their place. Every implemented change has an entry in the benefits register with its baseline, its counting rule and its audited value, with double counting struck out. It lets the sponsor say, with a number that survives a change of management, what the program has returned for every dollar it cost.
What you get
Two things come out of the program: a way of running maintenance and reliability that your own people operate, and the analysis that feeds it. The eight groups below are the first. They cover how work is identified, planned, scheduled and closed out, how its quality is audited and monitored, who owns each part, and how the people who run it are coached. They are what the asset manager and the maintenance manager own when the engagement ends, written in your system’s own terms and already in use. The analysis and the proof of value follow.
Frameworks you keep
The decision rules, written down once, so that the next manager inherits them.
- A criticality framework: assets scored on safety, environment, production and cost, ranked, agreed with operations and refreshed each year. It decides where maintenance and engineering effort goes.
- A tactic selection framework: for each failure mode, how the choice is made between condition monitoring, a fixed interval, redesign and run to failure, with the cost and risk test behind each choice.
- A loss framework: one downtime coding structure, a ranked top-ten list with jack-knife positions, and one named owner for each loss.
- An indicator framework: leading and lagging measures, each with a written definition, a counting rule, an owner and a target for each phase, reported on one page.
- A business case framework: every recommendation in dollars per operating hour or net present value with the driving assumption named, and a benefits register that counts each dollar once.
- A maturity assessment against the ladder above, with the evidence for each rung and the move to the next.
Roles and responsibilities made clear
Who owns what, in writing, before the work starts.
- A responsibility matrix showing who does the work, who decides, who is consulted and who is told, for the sponsor, asset manager, maintenance manager, superintendents, planners, schedulers, supervisors, tradespeople, the reliability engineer and operations.
- Role descriptions for the reliability engineer and the planner: what the job is, what it is not, how the week is split, and the time protected in writing.
- Decision rights: who may change a tactic, approve a business case, sign a management of change record and take a machine out of service, with dollar limits where they apply.
- The operations and maintenance interface: who agrees the weekly schedule, how access to plant is given, and what happens when production asks for a job to be moved.
- One owner and one date on every action, kept in a register that is reviewed in public.
- A staffing check of planners, supervisors and reliability roles against the ratios for your workforce.
Work management processes
The path of a job from request to close-out, written as it is run on your site.
- Work identification and priority: who may raise a work request, a priority system with response times, and a rule for what counts as emergency work, so that urgent work stops displacing planned work.
- A job plan standard: the tasks, parts, tools, labour hours by trade, permits, isolations and safety steps written for each planned job, and a planner’s check that a job is ready before it is scheduled.
- Scheduling: the weekly schedule built and locked, crews loaded to their available hours, and a break-in rule that says who may change the lock and records the reason each time.
- Execution and close-out: PM done as written within its window, and a close-out standard that records what failed, on which functional location, for how long and at what age, with findings and follow-up work raised before the work order closes.
- Backlog management: ready and not-ready work counted in crew-weeks, a review of aged work, and a rule for cancelling work that no longer applies.
- Parts and kitting: the rule for staging materials to planned jobs, and the check that the kit is complete before the job reaches the schedule.
Reliability processes and meetings
The processes that turn failures and data into changed plans.
- The meeting rhythm: the daily wall-walk, the weekly schedule, the weekly bad-actor review, the monthly reliability review, the quarterly strategy and business review and the annual plan. Each has a chair, a standing agenda, the inputs it needs and a decision it must produce.
- Root cause analysis: which failures trigger one, the method for each level of consequence, who is in the room, and how actions are closed and checked.
- Management of change for tactics: the task sheet, job plan and CMMS changed together and signed, so that a changed plan is a changed practice.
- Failure and data coding: the codes, the rule for recording component ages, and the enforcement at close-out that makes later analysis possible.
- The annual strategy review: criticality refreshed, tactics tested against the year’s failures, and the indicator set pruned and re-targeted.
Work quality and audits
How you know the work is done to standard: by sampling the records and the plant, not by asking.
- A monthly work order quality audit: a random sample of closed work orders scored against a written checklist (failure and component recorded, time and parts booked, readings entered, findings and follow-up raised), with scores by crew and by planner shown to the people who did the work.
- A planning quality audit: a sample of planned jobs checked for the tasks, parts, tools, labour hours, permits and isolations needed to do the job once, with no return trip to the store or the planner.
- A PM task quality audit: tasks read against the failure modes they are meant to catch, and the task sheets checked against what the crews actually do on the machine.
- A schedule adherence audit: every job on the locked schedule that was not done, and every job added after the lock, with the reason and who had the authority to break it.
- A wrench-time study: activity sampling of crews on shift at baseline and again at month 12, so that the time lost to waiting, travel, parts and permits is measured and not argued.
- A parts and kitting audit: jobs started without a complete kit, jobs delayed for parts, and the stock record for the critical spares checked against the shelf.
- A CMMS data audit: failure codes, functional locations and component ages checked against what happened on the machine, with the error rate reported each month.
- Each audit has a written method, a sample size, a scoring sheet, an owner and a date for the next one. Findings go in the action register with one owner and one date.
Monitoring and reporting
What is watched every week, who watches it, and what they must do when it moves.
- A weekly leading-indicator report: schedule compliance, PM compliance within the window, planned share of labour hours, backlog in crew-weeks, breaks in the schedule and close-out data quality, each with one owner.
- Limits and triggers for each indicator: the value that forces a response, the person who must respond and the time allowed.
- Data quality monitors: the share of work orders closed with a failure code, a component and a meter reading, reported weekly by crew, with a reason asked of the supervisor whose crew falls below the standard.
- Condition monitoring follow-up: every alarm logged with the action taken and the time from alarm to action, checked against the P–F interval for that failure mode.
- Bad-actor tracking: the top-ten losses re-ranked each month with the action, owner, date and the downtime moved since the action was closed.
- Root cause action tracking: the share of actions closed on time, and a check three months later that each closed action held.
- Post-change monitoring: after every tactic change, the failures, downtime and cost of that asset class are tracked, so that the benefit is measured and not assumed.
- A monthly management review pack: one page of indicators, the audit scores, the open actions and the decisions needed from the maintenance manager and the asset manager.
Coaching on your own plant
The people who will run the system are coached while the work is being done, on your own plant and data.
- Planners, schedulers and supervisors coached through the weekly cycle. Pardus sits in on the schedule build and the wall-walk, gives feedback the same day and checks the result against the compliance figures.
- The reliability engineer takes the first root cause analysis and the first Weibull fit alongside Pardus, co-leads the second and leads the third, with feedback after each.
- The maintenance manager and the asset manager coached to run the monthly and quarterly reviews: reading the indicator pack and the audit scores, challenging the numbers and turning findings into decisions.
- Crew briefings at each change on why the interval moved and what to do on finding something at the new limit. The test is whether a maintainer can say it in their own words.
Keeping it running
A system that depends on one person stops when that person leaves.
- Each process has a named owner, a version and a review date, and is held in one place that your people use.
- A way to change the processes themselves: how an audit finding or a better method alters a written process, with the change recorded.
- An annual audit of the whole work management system against the maturity ladder, by someone outside the maintenance department, with findings tracked to closure.
- A named deputy for each key role.
- The Phase 4 handover test: if the reliability engineer left, the reviews would still run the following quarter.
The analysis and the proof of value
- A ranked asset list, a loss profile with jack-knife positions, a data gap list and a baseline indicator pack inside 30 days.
- Root cause analyses that end in actions with one owner and one date each.
- Changed tactics loaded in your CMMS, with the task sheets and job plans changed with them.
- Weibull results with confidence bounds, intervals, spares holdings and life-cycle cost models in Excel comparing rebuild, replace, hire and run to failure, handed over unlocked, where the data supports them.
- Remaining useful life estimates from vibration and oil analysis data, where the data allows.
- A benefits register in dollars, audited by someone outside the program, and a 24-month roadmap against the ladder.
Site support
The program is run with your team and, where you want it, with a Pardus engineer on site. Site support is offered as part of the program, on its own, or to cover a gap in your reliability team. It is quoted on the days and the roster after a scoping discussion.
On site for the program
A Pardus reliability engineer on site on set days each week or fortnight, or in a block each month, to suit your roster and the phase. The days go to the schedule build, the wall-walk, the bad-actor review, the workshops with the crews and the audits. The analysis behind them is done between visits, remotely from your data.
Cover for the reliability engineer
Where the role is vacant, or the engineer is on leave or new, Pardus fills it for a period: the weekly bad-actor review, review of significant failures, checks of PM findings and condition monitoring calls, and the indicators and action register kept current. The handover is written down, so that the next engineer picks up live registers and not a blank page.
Mentoring on the job
A graduate or newly appointed reliability engineer, planner or scheduler works beside a Pardus engineer on real failures and real data: the first root cause analysis, the first Weibull fit, the first schedule build. Feedback is given on the day and the result is checked against the numbers.
Major failure investigation
After a significant failure a Pardus engineer attends site while the evidence is still there. The failed parts are kept and examined, the people who ran and repaired the machine are in the room, and the root cause analysis starts the same week. Actions leave with one owner and one date each.
Audit and review visits
A monthly or quarterly visit to run the work management audits, check the indicators against the records, walk the plant and brief the maintenance manager and the asset manager. It suits a site that runs the system itself and wants an outside check on whether it is holding.
Remote support between visits
Weekly report and data checks, analysis and business cases from your CMMS exports, and calls on Teams. A site that cannot host a visiting engineer on a roster can run the whole program this way, with site days kept for the workshops and the audits.
How the program is run
Who needs to be on the team?
A sponsor, usually the general manager or mine manager, who owns the mandate and the money. A maintenance manager who owns execution and the budget. Planners protected from today’s breakdowns, at about one per twenty to thirty maintainers. A reliability function of the order of two to four per cent of the maintenance workforce, which owns whether the work is the right work and owns no crews. Its numbers go to the sponsor unfiltered, because no function should be the only channel for measuring its own performance.
What does the reliability engineer spend a week on?
As a rule of thumb, about half on improvement work: analysis, tactics, root cause analysis and business cases. About a quarter supporting the maintenance process: reviewing significant failures, checking PM findings, closing out condition monitoring calls and keeping the data honest. About a quarter on communication: reporting, standing in front of crews and getting decisions made. Expediting parts, supervising jobs, owning the daily breakdown meeting and building dashboards nobody asked for are not the job.
Which meetings does it run on?
A daily wall-walk chaired by the superintendent, a weekly schedule build and lock, a weekly bad-actor review chaired by the reliability engineer, a monthly reliability review chaired by the maintenance manager, a quarterly strategy and business review chaired by the sponsor, and an annual plan. Each exists to make decisions; anything without a decision goes in the pack. Every action leaves with one owner and one date, tracked in public.
How does a recommendation become dollars?
Every recommendation arrives as dollars per operating hour or net present value, with the assumption that drives the answer named. Take a hydraulic cylinder with β = 2.03, a planned change-out at $42,200 and a failure at $258,200, which includes $18,000 an hour of lost production over a 12-hour repair. The optimum interval, near 3,500 hours, costs $24.42 an hour against $37.13 to run to failure, which is 34%. Across twenty excavators with four cylinders each at 6,000 hours a year, that is about $6.1 million a year. Cut the production loss behind the failure from $18,000 to $5,000 an hour and the optimum moves out to about 6,900 hours and the saving falls to 8%. The interval is not the lever; the consequence is.
How do you know a change has landed?
Ask a maintainer, not a supervisor, why the interval changed and what to do on finding something at the new limit. If the answer is in their own words and roughly right, it has landed. If the answer is “they changed it in the CMMS”, it is a document and not a change, and the tasks will drift back within two rosters. Changes go through management of change, and the task sheet, the job plan and the CMMS are changed together.
Why do reliability programs fail?
Rarely from a wrong answer. The common causes are tools bought before the basics, no usable data because change-outs were recorded against the machine and not the component, no owner and date beside each action, machine learning before PM compliance, a report that nobody implements, chasing certification instead of availability and cost, the reliability engineer becoming a spare supervisor, and a diagnosis that never ends. The day-90 gate exists to stop the last of these.
Start with what you already have
Reliability Engineering for Heavy Industry
Chapter 30 sets out this program in full, with the tables, the roles and the week-by-week plan.
Jack-knife diagram tool
Rank a year of downtime by what is frequent and slow to fix. The first analysis in the program, on data you already hold.
How to create a jack-knife diagram for downtime analysis
The method in Excel, and why it ranks losses better than a Pareto chart.
Root cause analysis for maintenance and reliability
Which failures to analyse, which method to use, and how to make the fix last.
Weibull calculator
When change-out ages exist, fit them in a minute, suspensions included.
Perform a Weibull analysis in Excel
The whole method, step by step, on hydraulic cylinder data.
Critical spares calculator
How many spares to hold for a target service level.
RAPTOR 7
Draw a reliability block diagram and simulate its availability.
Monte Carlo simulation in Excel for asset replacement
How uncertainty changes the overhaul, replace or rent decision.
Questions people ask
What is reliability engineering?
Reliability engineering uses failure, repair and cost data to predict how equipment will fail and to choose the maintenance, spares and replacement decisions that cost least over its life. On a mine or plant it answers questions such as which losses to fix first, when to change a component and whether to rebuild or replace a machine.
Our data is poor. Where do we start?
With the schedule and the close-out standard, not with analysis. The data health check in month one scores the five datasets the analysis depends on: work orders with failure codes, the downtime log, component change-out ages with suspensions, condition monitoring streams and cost by work order. Enforcing the standard when a work order is closed repairs them from then on. The jack-knife diagram runs on the downtime you already have, and Weibull waits until the ages exist.
Do we need to do Weibull analysis first?
No. Weibull analysis is useful once component change-out ages have been recorded, with the parts still running counted as suspensions. Before that, fitting it to whatever exists produces confident answers to the wrong question. The program puts it in Phase 3 and starts recording the ages in month one.
Should we buy condition monitoring or analytics first?
Not before the basics. A predictive alarm on a fleet at 60% schedule compliance produces a job that gets deferred anyway, and a model trained on a history in which half the breakdowns are coded as general repair learns that half of all failures are general repairs. Condition monitoring goes on the critical classes with a P–F interval worth having, as part of Phase 2, with limits set from baselines.
What happens in the first week?
Pre-start, then a walk through the workshop and the go-line with the maintenance supervisor, asking whoever is holding a spanner three questions: what breaks most, which job on the schedule is a waste of time, and what has been reported that is never fixed. Then four documents: the asset register with criticality if one exists, twelve months of downtime by asset and cause, the current PM schedules for the two most critical fleets, and last month’s PM and schedule compliance with their definitions. By the end of the week there is a top-ten loss list on one page, an honest rung on the ladder and one small thing fixed and visible.
What is a RAM analysis?
RAM stands for reliability, availability and maintainability. A RAM analysis models a system as blocks in series, parallel and standby, gives each block failure and repair distributions, and simulates it to predict availability and to show which items limit it. It sits in Phase 3, when the failure data supports it.
What data do you need?
To start: the asset register, twelve months of downtime by asset and cause, the PM schedules for the two most critical fleets, and last month’s PM and schedule compliance. For the Phase 3 analyses: work orders with failure codes and dates, component change-out ages, operating hours, downtime and costs, ideally two years or more, exported from your CMMS, with OEM parts prices and supply terms for the cost models.
Is the work remote or on site?
Both. The diagnosis starts on site, because the best-informed people about how a machine fails are the ones who repair and operate it, and the root cause and PM optimisation work is run as workshops with them. The analysis is done remotely from your data. Beyond that, a Pardus engineer can be on site on a roster for the whole program, or for a period of your choosing, as set out under site support above.
Can Pardus cover the reliability engineer role while we recruit?
Yes. A Pardus engineer can fill the role for a period, on site on set days and remotely between them. The weekly bad-actor review, the review of significant failures, the PM findings and condition monitoring calls, and the indicator pack and action register carry on, and the handover is written so that the next engineer inherits live registers. It is quoted on the days and the roster.
Can we commission one study instead of the whole program?
Yes. A single decision, such as rebuild or replace for one fleet, a spares review for one pool or a Weibull analysis of one failure mode, is a smaller job than a program, and about four weeks from data to decision. The data needs to exist; where it does not, the first step is to start recording it.
What does it cost?
It is quoted as a fixed fee after a scoping discussion, once the site, the question and the data are known. The 90-day diagnosis, a single study and a site-wide program are different sizes of job.
Who owns the models?
You do. The Excel models and the data are handed over unlocked, so your engineers can rerun them when prices or failure rates change.
Speak to the engineer who will do the work
Pardus scopes every project in a discussion with Miguel Pengel, the principal engineer, and a written scope with the deliverables, the fee basis and the dates follows. Larger scopes are resourced with named specialists or delivered with a partner firm.