Industries / Industrial equipment and services / IoT and industrial automation / Process instrumentation
Pressure and temperature
This work makes the sensors and transmitters that turn the pressure in a pipe or the temperature of a fluid into a signal a control system can read.
5
Milestones achieved, from 1886 to 1995.
1
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1886
Platinum wires measure heat
A thermocouple is two wires of different metals joined at one end, which give a small voltage that rises with the temperature of the joint. Henry Le Chatelier, a French chemist, published his tests of a pair made of pure platinum and platinum alloyed with 10% rhodium in 1886. NIST credits him with developing and testing this pair, now called type S, and says his work showed its reproducibility and stability at fairly high temperatures in air; it served as a standard instrument of the international temperature scale from 1927 until 1990. NIST's Monograph 175, issued in April 1993, gives the reference tables of voltage against temperature for the eight letter-designated thermocouple types in use.
Credit
Henry Le Chatelier; G. W. Burns and co-authors at NIST for the tables
1954
Silicon that senses strain
A strain gauge measures how much a part stretches from the change in a conductor's electrical resistance. Edward Simmons suggested the bonded metal strain gauge, a fine wire glued to the part, in 1936, and Arthur Ruge reinvented it in 1938. In 1954 C. S. Smith, visiting Bell Laboratories from Case Western Reserve University, made the first measurements of the "exceptionally large" change in the resistance of silicon and germanium under stress, an effect called piezoresistance. Silicon strain gauges more than fifty times as sensitive as metal ones followed; Kulite Semiconductor, founded in 1958, was the first licensee of Bell's piezoresistive patents, and in 1962 O. N. Tufte and colleagues at Honeywell Research built the first pressure-sensing diaphragm with the resistors diffused into the silicon itself. A 2009 review calls these sensors the first commercial devices that needed silicon machined in three dimensions, a forerunner of the micromachines (MEMS) inside today's sensors.
Credit
C. S. Smith (Bell Laboratories); Kulite Semiconductor; O. N. Tufte and colleagues (Honeywell Research)
1969
Capacitance replaces the lever
The first electronic pressure transmitters swapped the air-pressure signal of pneumatic instruments for a milliamp current but kept their force-balance mechanism, a set of levers and moving parts that drifted and needed upkeep. In 1969 Rosemount Engineering Co., which had made instruments for the US space program, introduced the Model 1151, which pairs a capacitive sensor, one that reads pressure as a change in electrical capacitance, with small electronics. Control magazine, interviewing Emerson, called it a step forward in stability, reliability and accuracy, in a package a third the size and weight of competing ones. An oil company's order for 1,000 units in 1972 made its name, and its successors, the digital 3051 about 20 years later and the 3051S in 2001, kept the line going.
Credit
Rosemount Engineering Co., now part of Emerson
1991
Pressure read as frequency
Yokogawa Electric's DPharp sensor reads pressure from the vibration of two tiny silicon resonators, made by micromachining and placed on a silicon diaphragm, one where the diaphragm stretches and one where it is squeezed. Pressure changes their resonant frequencies, which the transmitter measures; Yokogawa calls the method digital resonant sensing. It released the DPharp EJ series in 1991 and the EJA series in 1994, and said in 2012 that about five million EJA transmitters had been delivered.
Credit
Yokogawa Electric
1995
A thermowell breaks at Monju
A thermowell is the closed metal tube that reaches into a pipe to hold a temperature sensor, protecting it from the fluid. On 8 December 1995 one broke on the secondary sodium cooling loop of Monju, Japan's prototype fast breeder reactor at Tsuruga, and sodium leaked out; operators shut the reactor down by hand. Japan's Failure Knowledge Database records that the well failed by high-cycle fatigue from vibration in the direction of the flow, caused by vortices shed symmetrically behind it, while its designers, following ASME's 1974 code, had checked only vibration across the flow, and that a sharp step in the well's diameter concentrated the stress. The Japan Society of Mechanical Engineers issued a guideline on such vibration, JSME S 012, in 1998. ASME's PTC 19.3 TW-2010 replaced the 1974 calculation: it added this in-line resonance and, for liquids, cut the allowed ratio of vortex frequency to the well's natural frequency from 0.8 to 0.4, an Emerson engineer wrote.
LessonA design rule that checks only one way a part can fail leaves the others open, so the rule has to change when a failure shows a new one.
Credit
Power Reactor and Nuclear Fuel Development Corporation (operator); ASME's PTC 19.3 committee; JSME
Watch for
Clamp-on sensors match thermowells
A clamp-on (non-invasive) temperature sensor sits on the outside of a pipe and works out the fluid's temperature from the pipe's surface, so no hole or thermowell is needed. ABB says its NINVA sensors give "the same level of accuracy and performance" as invasive sensors, and in July 2026 it added Integrated, Remote and Compact versions. Omega's data sheet for its HANI clamp sensor claims ±0.5 °C on sanitary metal pipe and ±0.5 to ±1.0 °C on industrial pipe.
Achieved whenan outside laboratory or user publishes side-by-side readings against a sensor immersed in the flow, across flow rates, fluid temperatures and weather, with the data public.
Claimed by
ABB, July 2026
Open
01
Test clamp-on sensors in public
A published bench test of clamp-on temperature sensors from several makers, mounted on one stainless-steel water loop beside a reference thermometer immersed in the flow, run at a range of water temperatures, flow speeds, air temperatures and wind, with and without insulation on the pipe.
The briefCredit · unclaimed02
Recheck old thermowells in bulk
A program that reads a plant's thermowell data sheets and process line list, takes each well's dimensions and material and the fluid's speed, density and viscosity, runs the ASME PTC 19.3 TW calculation for every well, and lists those that fail the in-line resonance limit added in 2010, for an engineer to confirm.
The briefCredit · unclaimed
Open 01 · the brief
Test clamp-on sensors in public
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A published bench test of clamp-on temperature sensors from several makers, mounted on one stainless-steel water loop beside a reference thermometer immersed in the flow, run at a range of water temperatures, flow speeds, air temperatures and wind, with and without insulation on the pipe.
Why it is still open
The accuracy figures are makers' own: ABB says NINVA matches invasive sensors, and Omega gives ±0.5 to ±1.0 °C on industrial pipe. The only outside evaluation found is by Exera, a French association of instrument users, which lists a January 2025 evaluation of seven non-intrusive temperature sensors and a November 2025 good-practice guide for them among its reports for subscribers; no public results or raw data were found.
Built with what exists
A loop of 25 mm (1 inch) stainless-steel pipe with a small circulating pump and a thermostatic water heater kept below 70 °C; a calibrated platinum resistance thermometer and readout, borrowed from a calibration laboratory, immersed in the flow as the reference; clamp-on sensors such as ABB NINVA and Omega HANI bought at list price; a desk fan for wind; a data logger and Python.
First step
Build the loop with the reference thermometer and one clamp-on sensor, and log a week of readings while stepping the water temperature.
Done looks like
At least five products from at least four makers, each tested at five water temperatures from 20 °C to 65 °C, three flow speeds and two air conditions, with the error against the reference and the time to reach 63% of a step change published for every run, raw data included, and each maker sent its results for comment before publication.
Open 02 · the brief
Recheck old thermowells in bulk
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A program that reads a plant's thermowell data sheets and process line list, takes each well's dimensions and material and the fluid's speed, density and viscosity, runs the ASME PTC 19.3 TW calculation for every well, and lists those that fail the in-line resonance limit added in 2010, for an engineer to confirm.
Why it is still open
Wells sized under the 1974 rules were checked only for vibration across the flow, the kind of check that missed the Monju failure. The 2010 standard is not retroactive: wells already calculated under the 1974 rules need new calculations only when process conditions change. Makers offer calculators for their own wells, such as the free online tool an Emerson engineer described. A GitHub search for "thermowell" found 16 repositories and none that applies the standard's calculation to a list of wells; the nearest, ThermAssist (October 2026), checks a well's geometry against rules of thumb.
Built with what exists
A purchased copy of ASME PTC 19.3 TW; Python with NumPy; the open CoolProp library for fluid properties; an AI model that reads PDF data sheets and tables (see the Note "Seven more flagged" for a model finding problems in long documents that people had missed); worked examples from makers' calculators to check the code.
First step
Code the standard's natural-frequency and Strouhal-number steps (the Strouhal number sets how often vortices shed at a given flow) for one tapered well, and compare the results with a maker's online calculator at five flow speeds.
Done looks like
On 100 wells whose results were computed with a maker's calculator, the program's natural frequency within 2% and the same pass or fail for every well, then a run over at least 300 wells from a plant list shared with permission, reporting how many fail the 2010 in-line limit.