BRENNEN

Industries / Industrial equipment and services / IoT and industrial automation / Process instrumentation

Self-checks and safety ratings

This work builds checks into instruments so that they report their own faults, and rates them for safety instrumented systems, the separate systems that shut a process down when a reading crosses a limit.

4

Milestones achieved, from 2003 to 2018.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 2003

    Two routes into safety duty

    IEC 61511, the version of the functional safety standard IEC 61508 written for the process industries, came out in 2003. It grades each safety function by a safety integrity level (SIL); at SIL 2, in Yokogawa's summary, the function cuts the risk to between a hundredth and a thousandth of what it would be without it. For SIL 1 to SIL 3 it lets a plant put a transmitter or switch into a safety instrumented system by either of two routes: the device is designed and assessed under IEC 61508, or the user selects it on the basis of "prior use", with evidence that covers the maker's quality management, the device's performance in similar conditions and the volume of operating experience. For field devices, the standard says, that evidence is mainly the user's own list of approved equipment, kept up to date, with a device added only after enough experience and removed when it performs badly. Makers of assessed devices print the results: Emerson's Rosemount 2140 level switch, for example, is certified with a 97% safe failure fraction, the share of its failures that are either harmless or caught by its own checks, and 96% diagnostic coverage, the share of its dangerous failures that its own checks catch.

    law.resource.orgwebstore.iec.chyokogawa.comautomation.com

    Credit

    the IEC

  2. 2005

    Four signals for health

    Smart instruments report many fault codes of their own, and the codes differ from maker to maker. NAMUR's recommendation NE 107, first issued on 1 March 2005 and revised on 12 June 2006, sorts every diagnosis into four status signals that operators can act on: Failure, Function check, Out of specification and Maintenance required. NAMUR lists it as a partial replacement for NE 43 and issued its latest edition on 15 July 2025.

    namur.netnamur.netinstrumentationtools.com

    Credit

    NAMUR, with the German engineering bodies VDI/VDE and the user group WIB

  3. 2010

    Certificates meet field data

    A safety certificate's failure rates come from an analysis of the device's parts under reference conditions. SINTEF, the Norwegian research institute, wrote in the 2010 edition of its PDS data handbook, as quoted by the engineer Harvey T. Dearden in Control Global in 2012, that "vendor estimates of dangerous undetected failure rate are often an order of magnitude (or even lower) than that reported in generic data handbooks." It also found that random hardware faults typically make up only 30% to 50% of the dangerous undetected failures of sensors and valves, the rest coming from systematic causes, which Dearden ties to how devices are installed and kept. The 2021 edition rests on some thirty thousand maintenance notifications and work orders classified by hand, and SINTEF says the handbook is the preferred data source among Norwegian companies analysing safety systems.

    LessonA certificate describes the device, not the installation, so safety calculations need failure data from the field as well.

    controlglobal.comblog.sintef.com

    Credit

    SINTEF (PDS project); Harvey T. Dearden

  4. 2018

    A thermometer that self-calibrates

    In 2018 Endress+Hauser launched the iTHERM TrustSens TM371, which it calls the first self-calibrating thermometer. A reference material inside it has a Curie temperature, the point at which a magnetic material loses its magnetism, of 118 °C, and each time the process cools through that point the thermometer compares its platinum sensor with the reference and records the difference. In a case study at Merck Healthcare KGaA's sterile plant in Darmstadt, Germany, published in ISPE's Pharmaceutical Engineering in 2020, one ran for about four weeks in a steam sterilizer and made about 80 calibrations, all within specification, and calibrations at an accredited lab before and after showed no significant drift. The article reports that TÜV, a German inspection body, certified the calibration uncertainty at the reference point as ±0.35 K, and that TU Ilmenau analysed how a deviation at 118 °C bears on the rest of the range.

    miningweekly.comispe.org

    Credit

    Endress+Hauser; TU Ilmenau; Merck Healthcare KGaA (case study)

Watch for

  1. Safety data over HART

    FieldComm Group added SafeHART, a safety layer for the HART protocol, in HART revision 7.9 in May 2023, and says it is "designed to support Safety Integrity Level (SIL) 2 and 3 applications" over the same wiring and devices as ordinary HART, including WirelessHART and HART-IP. exida certified the protocol to IEC 61508 in August 2023, and FieldComm Group's ballot for HART 7.10, opened on 13 November 2025, revises the SafeHART requirements "based on developer feedback and recommendations from exida". No instrument or safety system using it was found on sale.

    Achieved whena transmitter and a safety system that use SafeHART are on sale and a plant or an outside assessor reports a safety function running on it.

    fieldcommgroup.orgsupport.fieldcommgroup.orgexida.comsupport.fieldcommgroup.org

    Claimed by

    FieldComm Group, November 2025

Open

  1. 01

    Score blocked-line detection

    A public bench test of the diagnostics that warn when an impulse line, the small pipe that carries pressure from a flow meter's orifice plate to its differential pressure transmitter, has clogged. Build a water loop with an orifice plate, fit transmitters and needle valves on the impulse lines to block them partly or fully on cue, change the flow and line pressure between runs, and score each method: the methods published by Waseda University and Yokogawa, written in open code, and any diagnostic built into a commercial transmitter.

    The brief
    Credit · unclaimed
  2. 02

    Coat the forks, count alerts

    An outside test of the self-diagnostics in vibrating fork level switches, which sense liquid by the change in a vibrating fork's frequency. Coat the forks with weighed layers of a soft material such as paraffin wax, step by step, and record at what thickness each switch's diagnostic warns of build-up and at what thickness it switches wrongly.

    The brief
    Credit · unclaimed

Open 01 · the brief

Score blocked-line detection

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

A public bench test of the diagnostics that warn when an impulse line, the small pipe that carries pressure from a flow meter's orifice plate to its differential pressure transmitter, has clogged. Build a water loop with an orifice plate, fit transmitters and needle valves on the impulse lines to block them partly or fully on cue, change the flow and line pressure between runs, and score each method: the methods published by Waseda University and Yokogawa, written in open code, and any diagnostic built into a commercial transmitter.

Why it is still open

The methods are known and their weak point is published. A Yokogawa patent notes that the pressure fluctuations used to spot clogging also change "depending on a flow rate as well as clogging of an impulse line," and a 2009 paper by Waseda University and Yokogawa researchers found that line pressure changes have "a harmful effect" on the diagnosis and that their phase-based method "remains to be improved." Each study used its authors' own rig. No published comparison of methods and products, with detection rates and false alarms under changing flow, was found.

patents.justia.comjstage.jst.go.jp

Built with what exists

a small pump, tank and 25 mm pipe loop running tap water below 3 bar; an orifice plate; two or three used differential pressure transmitters, including at least one sold with a blocked-line diagnostic; needle valves on both impulse lines; a fast data logger or a Raspberry Pi reading the transmitters over HART; Python with NumPy and SciPy for the fluctuation and phase methods.

  1. First step

    Build the loop with one transmitter and the two needle valves, and record ten minutes of pressure fluctuation with the valves open and with one closed most of the way.

  2. Done looks like

    At least 100 blockage events (partial and full, high side and low side) and at least 50 flow and pressure changes with no blockage, with the detection rate, time to detect and false alarms of every method published with the raw data.

Open 02 · the brief

Coat the forks, count alerts

Credit · unclaimed

The first to publish is credited on this line, and the deliverable moves from Open to Achieved.

What to build

An outside test of the self-diagnostics in vibrating fork level switches, which sense liquid by the change in a vibrating fork's frequency. Coat the forks with weighed layers of a soft material such as paraffin wax, step by step, and record at what thickness each switch's diagnostic warns of build-up and at what thickness it switches wrongly.

Why it is still open

The claims are makers' own. Emerson says the Frequency Profiling function of its Rosemount 2140 "immediately detects any build-up, fork blockage, or excessive corrosion," and publishes no threshold. A search of Crossref's index of published papers found no outside test of fork switch diagnostics.

automation.com

Built with what exists

a used Rosemount 2140 and fork switches from two other makers, such as Endress+Hauser's Liquiphant FTL51B; a water tank; paraffin wax, a scale and calipers to set and measure coatings; a USB HART modem with PACTware to read diagnostics; a camera to record each step.

  1. First step

    Buy one used Rosemount 2140, coat its fork in three thicknesses of wax, and log the diagnostic reading at each.

  2. Done looks like

    For each of at least three switches, at least ten coating thicknesses repeated five times, with the thickness at which the diagnostic warns and the thickness at which switching fails, published with photographs and logs.