Rotating Machinery Protection and API Standard 670

For those managing the reliability for large complexes such as refineries and petrochemical plants, protecting the rotating machinery that can number in the hundreds can be a daunting task.

Emerson's Gary Hawkins


Recently a refinery manager asked Emerson’s Gary Hawkins about options for protecting their 400+ rotating machines and the application of the API Standard 670 [$195 USD purchase price] in this effort.

Gary checked with Emerson’s Deane Horn for how this machinery protection standard has been applied. Deane explained that the API 670 is a vibration monitoring, rotating machinery shutdown protection specification and does not address seal systems such as API Standard 682 for pumps.

The typical requirements for a system to be in accordance with this purchasing specification are: 100 msec reaction time, redundant power source, redundant communications, shutdown voting logic, configurable for failsafe or availability, Modbus, accuracy specification, radial vibration, axial thrust position, temperature, case vibration, and remote relay bypass, to name a few of the key ones.

Deane noted that API Std 670 is a purchasing specification. Products cannot be certified and it is up to the end users whether to implement this level of machinery shutdown protection. API Std 670 can be applied to any rotating machinery including turbines, compressors, pumps, gearboxes, etc.

While API 670 can be applied to 400+ machines, usually predictive monitoring is a better strategy for this number of machines, since it is better to know issues ahead of time rather than to trip the machine and then start looking for the cause. Typically, process manufacturers and producers standardize on API 670 to shut down their large, expensive machines under dire conditions.

Examples of these situations are an impending bearing failure or overspeed condition, perhaps caused by a shaft coupling failure. Users can achieve significant benefit from monitoring the health of the machines before they have deteriorated to this point—where the trip condition, outlined by API Std 670, shuts down the machine. It’s best to think about the typical time to fail–is it milliseconds, violent, and dangerous? If so, then API Std 670 can be used for vibration monitoring and machinery protection.

The CSI 6500 machinery health monitor integrates protection, prediction, real-time performance monitoring and process automation

The CSI 6500 machinery health monitor integrates protection, prediction, real-time performance monitoring and process automation

On the other hand, if it is it hours, days, or months then predictive monitoring may be the right path. Predictive monitoring can be advantageous because plants are likely operating these machines to mechanical failure due to operating practices and lack the real-time feedback to the operators to know about the degradation of machinery health – information from predictive solutions means avoiding problems in the first place.

While API 670 has been around since the 1980s, the November 2014 update adds quite a bit of new content. Most notable is that there is an entire annex on machinery health monitoring and not just protection. In addition, there is an entire annex on wireless. The fifth edition grew to 256 page from 100 pages in the fourth edition and now addresses overspeed, surge detection, emergency shutdown systems, and annexes on prediction, wireless, reciprocating machines, and SIL examples to name a few.

If your process includes many rotating machines, make sure to get your hands on a copy of this standard and work with reliability experts with experience in applying the standard. You can also connect and interact with reliability and maintenance experts in the Reliability & Maintenance in the Emerson Exchange 365 community.

3 comments

  1. The reliability engineers have an arsenal of tools at their disposal to deal with the needs of different machines and equipment. In one end of the spectrum you can machinery protection systems with millisecond response time. In the other end of the spectrum you have portable vibration testers for weekly and monthly manual data collection. Somewhere in between you have wireless vibration transmitters to collect data maybe once an hour. All three solutions complement each other. See figure 1 and learn more from the article “Multi-parameter Condition Monitoring”:
    http://www.ceasiamag.com/2015/06/multi-parameter-condition-monitoring/

  2. Madni Ijaz says:

    Sir,

    I’m looking for the minimum requirements of API-670 for already installed machine, i.e. what is the least number of monitoring instrumentation that needs to be installed in order to declare it API-670 compliant.

    Looking forward to your kind response.

  3. Jim Cahill says:

    Hi Madni, Thanks for your question. I checked with friends in our machinery health area of this business and here’s what I heard back:

    API 670 is a purchasing guideline for the shutdown protection system for critical turbomachinery assets. The compliance or alignment with API 670 is typically decided by the end user except, in some cases a third party, such as an insurance company or an internal specification or an asset warranty may drive the need for API 670 alignment. If there is an insurance, internal or warranty driven specification then these specifications will typically dictate what the expected coverage should be otherwise there is significant leeway to decide what is acceptable.

    So getting back to the actual question; without details it’s very hard to answer the question specifically. It would help to know if a specific asset is being considered and how critical this asset is to the production process. It would help to know what sort of bearings this asset has, sleeve or antifriction. Does the asset involve a steam or combustion turbine where it would need overspeed monitoring and thrust protection.

    Plus keep in mind that for the asset to align with the API670 standard both the sensors and protection system have to align with the standard. So my best answer to your original question is to contact the asset vendor and ask them what their shutdown protection recommendations are. If an existing asset does not have a protection system already implemented then that to me indicates it’s not a highly critical asset.

Leave a Reply