BRENNEN

Industries / Industrial equipment and services / Industrial services / Pump and valve repair

Testing

Testing is how a shop or plant proves that a repaired pump or valve works and does not leak: performance runs that measure flow, pressure and power, pressure tests of valve bodies, seats and packing, set-pressure checks of relief valves, and the leak surveys that plants make in service.

6

Milestones achieved, from 1983 to 2024.

1

To watch for: announced, not yet confirmed outside.

2

Open deliverables. Nobody has shipped them.

The road

AchievedWatch forOpen

Achieved

  1. 1983

    A leak defined in ppm

    On 18 October 1983 the US Environmental Protection Agency (EPA) set standards for equipment leaks at new synthetic organic chemical plants (40 CFR Part 60, subpart VV). Pumps handling volatile liquids must be checked monthly with a portable instrument under EPA Method 21, which samples the air at the seal for organic vapor, and weekly by eye for liquid dripping from the seal; a reading of 10,000 parts per million (ppm) or more is a leak. A first attempt at repair, such as tightening the packing gland nuts or making sure the seal flush is running at its design pressure and temperature, is due within 5 calendar days and the repair within 15, and pumps fitted with dual mechanical seals and a barrier fluid system are exempt from the monthly check.

    law.cornell.edu

    Credit

    US EPA

  2. 1998

    Relief valves tested in place

    In-place testing checks the set pressure of a spring-loaded relief valve without removing it, by lifting the spindle with a measured force while the plant runs. TEAM, which offers the method as Trevitest, says the service was first developed in 1979 and claims accuracy within 0.5%. Researchers at the University of Pretoria followed three relief valves at Sasol's plant in Secunda, South Africa, through removal, transport, repair and reinstallation, and found Trevi readings repeatable within 0.55% and changes between stages of at most about 3%. Compared with Sasol's bench tests of overhauled valves, Trevi readings could be relied on only for set pressures of 560 kPa (81 psi) and above. Sasol had found that many valves began to leak at lower pressures than their setting; the authors concluded that the spring setting did not physically change and that the valves could still leak because of poor seating.

    scielo.org.zateaminc.com

    Credit

    A. M. Muller and M. Heyns, University of Pretoria, with Sasol

  3. 1999

    Audits find hidden leaks

    EPA's National Enforcement Investigations Center (NEIC) monitored valves itself at 17 petroleum refineries and compared its results with the refineries' own Method 21 records. The refineries had reported 2,266 leaks among 170,717 valves, a leak rate of 1.3%; NEIC found 2,372 leaks among 47,526 valves, 5.0%. EPA, which explained the results in a 1999 enforcement alert, put the gap down to monitoring staff not following Method 21 and to weak quality control, and the guide adds that inspectors have found a data logger's time stamps showing valves "monitored" at one per second.

    LessonA leak test is only as good as the way it was done, so an audit has to repeat the measurement, not just read the log.

    epa.gov

    Credit

    EPA NEIC; summarized in EPA's 2007 best practices guide

  4. 2002

    Tested before and after repair

    California's 2001 energy crisis led to a state-funded program for testing irrigation pumps. The Advanced Pumping Efficiency Program (APEP), run by the Center for Irrigation Technology at California State University, Fresno, subsidized 49,925 pump efficiency tests from 2002 through 2017 and paid 2,875 rebates for pump overhauls and retrofits, which required a test before and after the work. A test measures flow, discharge pressure, well lift and power to estimate the pumping plant's overall efficiency. In the University of California's journal California Agriculture in 2002, Blaine R. Hanson drew on 20 years of tests of pumping plants in the San Joaquin Valley. A summary of 63 pumps tested before and after repair showed capacity up 39% and overall efficiency up 33% on average, but the motors drew more power, so a pump run for the same hours as before used about 17% more energy, and only pumping the same volume of water as before saved energy, about 22%. He noted that operators commonly run repaired pumps for the same hours as before.

    LessonJudge a repair by the energy used per unit of water over a season, not by the efficiency figure from one test.

    pumpefficiency.orgpumpefficiency.orgescholarship.org

    Credit

    APEP and the Center for Irrigation Technology; Blaine R. Hanson, University of California, Davis

  5. 2014

    Valves type-tested for leaks

    API 624, issued by the American Petroleum Institute in February 2014, type-tests a design of rising-stem valve with flexible graphite packing: a sample valve is opened and shut 310 times through three heat cycles from room temperature to 260°C (500°F), at a test pressure of up to 600 psig (41 bar), and leakage at the stem must stay at or below 100 parts per million by volume (ppmv), with no tightening of the packing allowed. The packing must first pass API 622, a test of the packing alone over 1,510 cycles, whose third edition also caps leakage at 100 ppmv with no gland adjustment. ISO 15848-1, issued in January 2006, sets an international scheme, and API 641, issued in October 2016, covers quarter-turn valves.

    valvemagazine.comvalvemagazine.compumpsandsystems.com

    Credit

    American Petroleum Institute; ISO

  6. 2024

    Gas cameras get rules

    Appendix K to 40 CFR Part 60, published by EPA on 8 March 2024, is the protocol for finding leaks from valves, pumps, seals, flanges and other equipment with optical gas imaging (OGI) cameras, infrared cameras that show hydrocarbon vapor as a visible plume. Each camera's detection limit must first be established across an "operating envelope" of conditions and settings; operators must image each component from at least two angles, for at least 10 seconds per angle in a scene of 10 or fewer components, and should not survey for more than 30 minutes without a break. The protocol is for finding leaks, not for measuring how much leaks.

    law.cornell.edu

    Credit

    US EPA

Watch for

  1. Robots look for gas leaks

    ADNOC says a heavy-duty inspection robot made by Taurob of Austria, fitted with 3D lidar and thermal cameras, is making autonomous rounds at its Taweelah gas compression plant in the UAE so that engineers can identify potential gas leaks and hot spots, and that an "operator" robot able to turn valves and read gauges, developed in the ARGOS joint industry project with Equinor, Petrobras, TotalEnergies and others, is due to be working by the end of 2026. The release names thermal cameras, not gas imaging cameras, and gives no leaks found.

    Achieved whenADNOC or an outside party publishes the leaks the robot found compared with an OGI or Method 21 survey of the same equipment.

    hydrocarbonengineering.com

    Claimed by

    ADNOC and Taurob, May 2026

Open

  1. 01

    Did the repair save energy

    A public table of pump repairs measured before and after, built from tests that subsidy programs already required: for each job, the work done (impeller adjustment, repair of a worn pump, replacement of a mismatched pump), the flow, head, kilowatts and overall efficiency before and after, and, where owners share meter data, the kilowatt-hours per unit of water pumped in the following season.

    The brief
    Credit · unclaimed
  2. 02

    Count the bubbles by camera

    An open camera counter for the bubble test used to judge whether a closed valve seat or a relief valve seat seals: gas that leaks past the seat is led through a small tube under water, and the bubbles are counted for a minute against a limit. The counter would count the bubbles from video, estimate their volume and save the clip with the result.

    The brief
    Credit · unclaimed

Open 01 · the brief

Did the repair save energy

Credit · unclaimed

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

What to build

A public table of pump repairs measured before and after, built from tests that subsidy programs already required: for each job, the work done (impeller adjustment, repair of a worn pump, replacement of a mismatched pump), the flow, head, kilowatts and overall efficiency before and after, and, where owners share meter data, the kilowatt-hours per unit of water pumped in the following season.

Why it is still open

The tests exist but the results are not public. APEP alone subsidized 49,925 pump tests in California from 2002 through 2017 and required tests before and after for its 2,875 rebates, but its site reports counts and dollars, not what each repair changed. The largest pooled figure found is older: Hanson's 2002 paper cites his own 1988 summary of 63 pumps tested before and after repair. A web and Crossref search found utilities' single-site case stories but no public analysis of the thousands of tests made since.

pumpefficiency.orgescholarship.org

Built with what exists

Anonymized test reports from a program office or from pump test companies; a model that answers the same fixed questions about each scanned report, such as Gemini Decision 1 (see the Note "A thousand questions, one pass"); pandas or DuckDB for the table; utility interval meter data where owners agree.

  1. First step

    Ask the Center for Irrigation Technology at Fresno State whether anonymized before-and-after pairs can be shared, and code 30 reports from one test company by hand to settle the fields.

  2. Done looks like

    At least 300 before-and-after pairs coded, with a second person checking 50 of them; published medians and ranges of the change in overall efficiency, flow and kilowatts for each kind of repair, and kilowatt-hours per acre-foot (about 1,233 cubic meters) for the jobs with meter data.

Open 02 · the brief

Count the bubbles by camera

Credit · unclaimed

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

What to build

An open camera counter for the bubble test used to judge whether a closed valve seat or a relief valve seat seals: gas that leaks past the seat is led through a small tube under water, and the bubbles are counted for a minute against a limit. The counter would count the bubbles from video, estimate their volume and save the clip with the result.

Why it is still open

In the Pretoria study of Sasol's relief valves, the bench test lowered the pressure to 90% of the set pressure and counted the bubbles, with no camera named. G Systems of Texas sells a counter for this test, compatible with API 527 (the relief valve seat tightness standard), that shows bubble count, inferred flow and camera images, but its 2016 release gives no accuracy figures. R. C. Hart and co-authors (2023) compared software counts with human counts for underwater video of oil and gas seeps, not valve tests, and simple OpenCV bubble counters for recorded video are on GitHub, with no calibration against a measured flow. No published comparison of a camera counter with measured gas flow in seat tests was found.

scielo.org.zaprnewswire.comapi.semanticscholar.orggithub.com

Built with what exists

A Raspberry Pi Camera Module 3 or a phone; OpenCV blob tracking; a bench rig of shop air regulated to no more than 1.7 bar (25 psi), a needle valve, a standard tube in a jar of water, and a soap-film flowmeter to measure the true gas flow.

  1. First step

    Build the needle valve and tube rig on a bench and record 20 minutes of video at five flow settings.

  2. Done looks like

    200 one-minute runs from 0 to 60 bubbles a minute; software counts within 1 bubble a minute of the average of two people counting the same videos frame by frame on 95% of runs; estimated flow within 20% of the flowmeter; rig design, videos and code published.