Home > FAQ

Frequently-asked Questions (FAQ):

The most frequent questions being raised during our day-to-day work, on company visits, events and conferences are included in our list of 24 FAQ displayed on this page. These questions address seven topics:

If you have questions not included in the list displayed, we invite you to contact us. You would give us a valuable opportunity to complete our FAQ list.

Any burning questions? Contact Us

The most frequent questions being raised on company visits, events and conferences are included in our list of 24 FAQ displayed below. If you have questions not included in this list, we invite you to contact us and to give us an opportunity to complete our FAQ list.

General questions

  • What do I get for my money? How does the end product look like?
  • As a first step, we configure the Cassantec solution to your particular asset and the malfunctions that are typically related. Together with you we set up a list with typical malfunctions that occur and the risk tolerance level for each one of them.
  • After the configuration process you will receive periodical updates of the prognostic report. The report contains the probability that certain malfunctions occur in the future. Based on risk tolerance levels which we established with you, you will immediately see which components are likely to become a problem at what point in time in the future.
  • You can access the report on an online portal. You can get an overview at the fleet level view and go more into detail to specific components and their related malfunctions and risks.
  • We offer you a live online demonstration of the report, so that you can get to know all its advantages and possibilities. Please do not hesitate to contact us for more information or to make an appointment!
  • Does Cassantec provide earlier warnings and better diagnostics than traditional asset management software?
  • Cassantec's unique proprietary technology is proven to be highly accurate and reliable. The prognostic foresight is unique and clearly offers more advantages than basic diagnostic tools.
  • We are already pretty advanced in condition monitoring. Can you provide us with anything new?
  • Well-established condition monitoring processes are actually a prerequisite for the effective use of our Prognostic Report. And on the basis of advanced condition monitoring capabilities, we do indeed provide something new: fundamentally new insights from condition diagnostics and malfunction prognostics allow us to further leverage your benefits from the condition monitoring efforts already spent.
  • What is your solution able to do that others can´t? What is the difference to other asset management tools?
  • Prognotics
  • … Versus Condition Monitoring Software:
  • We relate the condition data to the current and future malfunction risks whereas condition monitoring software is based on data projections, illustrations, and comparisons to generic alarm levels.
  • … Versus Diagnostics and Predictive Analytics:
  • We answer the questions related to timing (when) whereas Diagnostics and Predictive Analytics focus on answering questions on the current equipment condition (what / why / where).
  • … Versus ERP and Asset Management Software:
  • We provide decision support based on the current and future asset conditions and malfunction risks whereas ERP and asset management software offer limited top-level statistics.
  • … Versus Work Order Management Software:
  • We help prioritize and select work orders based on malfunction risks for critical assets. This, in turn, may be important input for work order management.
  • Prognosis vs. diagnostics, what is the difference? What are typical questions that can be addressed?
  • We answer the questions related to timing (when) whereas Diagnostics and Predictive Analytics focus on answering questions on the current equipment condition (what / why / where).
  • Questions addressed by Diagnostics are:
  • What is the condition of our power plant?
  • Why is this condition critical, or why not?
  • Which parameters are most indicative?
  • Where are the problem’s root causes?
  • How do we best resolve challenges?
  • Questions addressed by Prognostics are:
  • When will this condition become critical?
  • When will we get a warning, alert or alarm?
  • When will be the best time to fix problems?
  • Will we make it until the next scheduled outage?
  • Will other plants in the fleet have the same issue?
  • Could you give us a general diagnostic view of our asset, without the prognostics?
  • Yes, of course. How long should this diagnostic view be valid? We are immediately back to prognostics…
  • We are already using for condition monitoring, reliability management, and/ maintenance support. What can you provide in addition?
  • We differ from ERP (e.g. SAP) and asset management software (e.g. Meridium) in that we provide maintenance decision support based on the current and future asset conditions and malfunction risks. This goes far beyond the limited top-level statistics available to ERP and offers additional and powerful statistical modeling capabilities to asset management software.
  • We differ from work order management software (e.g. IBM Maximo) in that we recommend work order selections based on malfunction risks for critical assets. This, in turn, may be critical input for work order management.
  • We differ from condition monitoring software (e.g. Emerson) in that we relate the condition data to malfunction risks at present and over different future time horizons. This goes far beyond the diagnostic and prognostic capabilities of condition monitoring software, which is based on data projections, illustrations, and comparisons to generic alarm levels.

Application Process

  • For which types of industrial assets can you provide a Prognostic Report?
  • At this point, we are looking at fast rotating industrial assets such as turbines, pumps, compressors, some large engines, and gearboxes. These types of assets often have similar critical components, such as bearings, which determine the overall asset reliability at system level. In the future, we may target further asset types, such as electrical or hydraulic assets, with our Prognostic Report.
  • For the specific asset to be reported on, three criteria should be met: The asset should (1) be critical for safe plant operation, (2) involve considerable costs and effort of unscheduled downtime, and (3) generate condition and process data that is archived and available via electronic download.
  • Which asset failure modes does your Prognostic Report consider?
  • Following the Pareto observation, a small fraction of all possible malfunction modes is typically responsible for the bulk of unscheduled maintenance risk. Hence, we select the top 12 asset-condition-related malfunctions that can be anticipated with knowledge of the asset's condition. Clearly, there will always be a small residual risk of a malfunction that we have not considered or that we could never anticipate.
  • The set of selected malfunctions very much depends on the asset, on its environment, and on the condition data available. In particular, three criteria should be met: The specified malfunctions should (1) address the dominant technical challenges of the selected asset relating to temperature, vibration, or lubrication, (2) be essentially detectable before asset failure or damage occurs, and (3) be as independent of each other as possible.

Data questions

  • What condition data do you need for your Prognostic Report?
  • At this point, we are looking at lubricant, vibration, and thermal condition parameters at the asset component level. The asset condition parameters are sometimes specifically defined for the asset analyzed, and receive high attention from condition experts in daily operations. Many of these condition parameters are actually functions of absolute or relative data measurements, time, and other parameters. We prioritize condition parameters depending on their dynamics (variability of values, speed of uptake and downtake) and on their relation to the malfunctions considered.
  • The defined condition parameters are generally subject to three criteria: They must (1) be arithmetic or logical functions of the available condition data, process data, and time, (2) relate to at least one of the malfunctions specified, and (3) be well defined for a probabilistic assessment, i.e. estimation of probabilities of reaching normal, advanced, significant, or extreme value level given one of the specified malfunctions.
  • Our approach is open to inclusion of new types of condition data, such as filter debris analysis or stress wave analysis.
  • What process data do you need for your Prognostic Report?
  • We typically use the asset's speed, load, or utilization data, in order to verify the validity of condition data and to scale certain model parameters.
  • How much data is typically necessary to make the outcomes reliable and useful?
  • Typically, 3 to 5 years of data history is ideal and yields a very strong outcome. Yet, also with shorter data histories good outcomes can be achieved.
  • What is a useful time interval in which the data should be updated? Conditions can change rapidly, how do you take charge of that? If our report is a week old and does not contain most recent developments, isn´t it already obsolete then?
  • The time interval depends a lot on your preference and the purpose you are using the reports for. For a periodical update about the Remaining Useful Life (RUL) for instance, a weekly update seems decent. The reports do not incorporate real-time data. Accordingly, events that occur after we got the data from you are not considered. In general, the more up-to-date you want the reports, the shorter the time intervals in which the data is updated should be.
  • We are taking condition data, but we do not archive it. Can you generate forecasts for our asset without any historical data?
  • Yes, in some cases. First of all, we may have access to suitable historical reference data on comparable assets. Without such external reference data, we would first have to gather a representative set of condition data from your asset over a certain period of time before providing conclusive Prognostic Reports.
  • We are not taking condition data at all. Can you generate forecasts for our asset without any condition data?
  • Unfortunately, we can not. But if your asset is truly critical to your operations, we recommend you look into condition monitoring!

Technology Details

  • What is the statistical/mathematical method behind the technology you are using?
  • Our patented technology consists of two sequential steps: 1. Data trends are identified and prognosticated into the future. The main methodology used for this step is Markov chains. 2. These particular trends and their prognoses are then related to certain malfunctions. The main methodology used for this step are Bayesian Networks.
  • As a consequence our solution yields a much more precise and accurate outcome, as compared to using regression analysis, data mining, Monte Carlo Simulation and the like. The accuracy of the solution has been proven in several retro perspective projects. 99% of malfunctions were predicted accurately by Cassantec.
  • What precisely do you read out of our asset condition data?
  • Loosely speaking, we determine (1) where the overall asset condition is standing, (2) in what direction it is heading, and (3) how fast it is moving. Formally speaking, we analyze the asset's parametric condition and the first and second derivatives of the asset's condition development over time.
  • In addition, we determine the asset's steady state, a parametric condition in which the asset is most stable from an abstract statistical perspective. The proximity of the asset's current condition to its steady state can be indicative for latent technical challenges that cannot be inferred from the most recent condition data sample alone.
  • The aforementioned analytical tasks require current and historical condition and process data. The prognostic inferences achieved go far beyond the scope of classic condition monitoring, and are not included in the standard diagnostic offerings on the market.
  • Don't you compare vibration and lubricant condition data like apples and oranges?
  • Essentially, we do not compare apples and oranges, but calories and calories.
  • We do not directly compare data from different condition monitoring technologies. A comparative analysis is done only after condition data has been related to malfunction risk. The risk of malfunctions, in turn, can be compared.
  • How can you predict asset failure without failure data?
  • We do not predict failure – we predict malfunctions based on condition data.
  • An asset failure corresponds to a breakdown, typically leading to unscheduled maintenance. Such breakdown is usually the result of a long period of accumulating damage. Repair, down time, and foregone product are typically too expensive to accept a breakdown. Asset failure may also include life, health, and environmental hazards. Hence, we have limited explicit failure data in the first place.
  • A malfunction, on the other hand, is an abnormal operating condition that, if not treated, may cause degradation in performance, a failure, and/or an unplanned shutdown. By definition, a malfunction indicates the end of the asset's remaining useful life (RUL), although the asset may still be useful. We want to detect malfunctions as early as possible to schedule maintenance, repairs, or process adjustments. We want to compute the probabilities of malfunctions given condition data, assuming there is a connection between asset malfunction and condition. Clearly, not all malfunctions can be predicted based on the asset's condition.
  • We are sometimes asked to use the term failure mode rather than malfunction mode, as it corresponds to reliability management terminology, yet we believe that for technological progress in predictive maintenance (PdM), a distinction between failure and malfunction is needed.
  • How far into the future can you predict asset malfunctions?
  • The prognostic horizon that we can offer depends (1) on the asset, (2) on the malfunctions considered, and (3) on type and quality of the condition data.
  • Some malfunctions allow a very long prognostic horizon. For instance, some assets in the mining industry are subject to linear abrasive corrosion. Knowledge about the surface coating of the corresponding asset components and about the overall asset utilization allows predicting the corrosion-related malfunctions with high accuracy over years. Other malfunctions, such as that of a rotor rub in a turbine generator of the power industry, may appear more rapidly without much possibility of prediction via condition data.
  • The prognostic horizon of our Prognostic Report can often be extended by collecting different or additional condition data, possibly from new sources or through new technologies. Such possibilities of condition data enhancement are explored with the operator during the Prognositc Report configuration process.
  • You provide some of the information that we need in the form of a probability distribution. We don't really like probabilities. We prefer facts. Could you avoid probabilities?
  • Since the risk of malfunction is inherent in the operation of your industrial asset, we are not able to completely avoid this risk. Hence, we are bound to use probability figures or distribution curves to communicate risk-related diagnostic insight and prognostic foresight to you. In so doing, we can help you to better analyze, manage, and reduce this risk, and to get closer to the facts!

Special Applications

  • We are operating a hydropower plant. Can you help us to predict and reduce the abrasive corrosion on the coating of our hydro turbine blades?
  • Yes, of course we can. We actually already have a successful project with an hydropower plant operator running.
  • We are operating a wind park. Can you provide Prognostic Reports for our wind turbines and gearboxes?
  • Essentially yes, depending on what condition data you are taking and when, where, and how this data is downloaded, transferred, and archived. Please contact us to discuss this.
  • Can you provide a periodical Prognostic Report on non-operating, redundant assets that we keep for safety purposes? Maybe for an emergency diesel generator?
  • Unfortunately, we cannot. We require operating condition and process data from your asset. Non-operating assets do not provide the condition information that we need for accurate prognostics. Chances are good that malfunctions caused by limited operation of the asset require very different condition data sources for diagnostics and prognostics.

Market Relations

  • Why isn't your Prognostic Report offered by an asset manufacturer, by a maintenance service provider, or by some other established player in the maintenance market?
  • Most established players in the original asset and maintenance services market benefit from the prevailing preventive maintenance solutions, based on legacy processes and tools. They benefit from the asset operators' shortness of predictive information, leading to higher average maintenance budgets spent. Hence, there are limited incentives to move forward to predictive offerings.
  • Beyond shortness of incentives, there is an obvious technological challenge: it is anything but trivial to generate high-quality prognostic reports for critical assets based on the available condition and process data. In fact, it took us a few years to make this work.
  • Hence, the more established candidates have not come up with a similar offering over the last decades. A credible predictive offering would have to come from a relatively young firm with demonstrated technology profile, strong drive for progress, and high enthusiasm and a clear business perspective for serving the operator's interest through a new type of prognostic service offering. That is what Cassantec is all about…
  • Asset manufacturers have the most knowledge about their assets. Why don't you offer this Prognostic Report in joint venture or collaboration with the manufacturer?
  • Many asset operators would disagree that after operating and maintaining their asset for decades, they know less about it than the manufacturers. Furthermore, the operator's job is to sustain asset operation while balancing costs of preventive maintenance measures with expected costs of unscheduled maintenance. The business objective of an asset manufacturer after warranty expiration is to sell new assets, to provide early upgrades for installed assets, to maximize profits in the aftermarket, and to minimize liabilities from all of the above. Our Prognostic Report favors the operators' job more than the manufacturer's objective. That is why we prefer collaborating with the operator.
  • Yet, on special occasions, we do offer a computational solution to select asset manufacturers as a complement for their aftermarket service product portfolio.
  • Why wouldn't lubricant or vibration analysts offer this type of prognostic (RUL) information in addition do their diagnostic services?
  • They certainly could. In fact, we are collaborating intensively with lubricant and vibration experts in the context of configurations of our Prognostic Report. Condition experts do appreciate our offering.
  • From the methodology and technology perspectives, however, condition experts are used to work with relatively simple, rule-based systems, applying experience-based alarm limits to recent condition parameters. There is no advanced predictive model that would go anywhere beyond linear regression. Hence, our Prognostic Report would not be part of condition analysts' value proposition.
  • We do not believe that this will change in the near future. Innovation in condition monitoring is targeting robust data generation and comprehensive, versatile service offerings. Advanced prognostic models are not really on the innovation agenda.

Result Verification

  • How can we verify the accuracy of the reports that you have provided?
  • You can verify the results through (1) a formal concept validation, (2) plausibility checks, and (3) a practice review. In fact, we apply these techniques ourselves to verify the results we deliver.
  • (1) The formal concept validation involves a one-time methodology examination by an official expert panel, as well as academic and professional endorsements. It also involves a transparent, continuous formal monitoring of our prognostic solution in every industrial application (statistical significance of data, applicability of data sources, etc.).
  • (2) The plausibility checks (ex-ante) involve the periodical monitoring of the Prognostic Report's plausibility by your trusted in-house asset condition experts. Such plausibility checks address in particular measured and assessed data (repot input) as well as quantitative results, interpretations, and recommendations (report output). Furthermore, Prognostic Reports may be subject to sporadic peer reviews and post audits by external experts.
  • (3) The practice review (ex-post) involves a periodical check whether prognoses have or have not materialized (when, where, how, why), and may trigger a feedback process addressing improved data sourcing or revised condition parameter assessments. Practice reviews also consider checks whether equivalent prognoses have or have not materialized in comparable equivalent circumstances.
  • Has your approach ever been checked by independent experts?
  • Yes, an independent expert council of the Swiss government's CTI program has audited and validated our approach. If you consider working with Cassantec and this is a point of concern, we'll be happy to get you in touch with the CTI coordinator in charge of the expert audit.

Know-how Confidentiality

  • Do you sell off the know-how that our in-house condition experts bring in?
  • No, your know-how and data will always be protected: every Prognostic Report has its own set of parameters, which we determine and record once in the beginning the application. We do not transfer this data between operators. Besides, this would be very difficult, since the asset, condition monitoring processes, and reliability standards differ widely between asset operators.