TESTING OF NUMERICAL RELAYS
TESTING OF NUMERICAL RELAYS
(OP)
I would like to consider again this issue:
What does testing a numerical relay mean? Why do we do it?
What happens if the relay fails a test?
Which kind of tests are required for digital relays and what is their purpose?
(for example: function test like overcurrent or distance protection tests; system test like communication schemes, A/R logics; maintenance test (purpose? )
Please let's try to figure out what we are asking and what we are expecting form a test of a digital relay. From scratch. I think that not even the relay manufacturers have a clear idea about that!
We should maybe start to list (that is easier) the arguments/needs/goals for the testing of the electromechanical relays, than remove what is not needed anymore because of technological change, and see what is new and needs to be tested, and why.
Try to make reference to your experience/troubles. Find out situations that have been detected thanks of the test.
Let's see if we manage this
What does testing a numerical relay mean? Why do we do it?
What happens if the relay fails a test?
Which kind of tests are required for digital relays and what is their purpose?
(for example: function test like overcurrent or distance protection tests; system test like communication schemes, A/R logics; maintenance test (purpose? )
Please let's try to figure out what we are asking and what we are expecting form a test of a digital relay. From scratch. I think that not even the relay manufacturers have a clear idea about that!
We should maybe start to list (that is easier) the arguments/needs/goals for the testing of the electromechanical relays, than remove what is not needed anymore because of technological change, and see what is new and needs to be tested, and why.
Try to make reference to your experience/troubles. Find out situations that have been detected thanks of the test.
Let's see if we manage this






RE: TESTING OF NUMERICAL RELAYS
1. We also take in account digital relays, becouse I'm always separate digital and numerical relay.
2. Test of relays at commissioning time and at maintanance test.
3.Commissioning time, full test:
3.1 Analog inputs
3.2 Digital inputs ( include threshold voltage operation), outputs
3.3 Settings
3.4 Logics
3.5 power supply ( min and max thresholds).
4. Maintanence test, only 3.1, 3.2 and 3.5 (is very important test, electrolit capacitors into PS, off/on PS test).
Maintanence test depend on : relay is connected to SCADA or not, if yes pereodical test once per 6-7 years, if not per 5-6 years. Not needed test of settings and logics, only inputs and outputs. IRF/watchdog/selfsupervision not controlled Analog inputs ( I'm not say about AD converters), BI/O. I'm also recommend provide megger 500V to analog inputs.
OK, it's only start of topic.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
Divide maintenance to periodic testing and preventive maintenance. Periodic testing for example per 3-4 years: what are include:
1. Visual inspection.
2. Front panel inspection.
3. Communication inspection.
4. Self-supervision inspection.
If you have SCADA, you can check many parameters via event/parameter list, compire settings with original, meas. with some power meters.
Preventive test: why is important, you have two groups of components: electronics with some life cycle influenced by themperature, hummidity,RF, etc and electromechanical with other life cycle influenced by corrsion, vibration, duct, etc. will be provide each 8-10 years.
Of course,preventive maint. is include CT/VT, swg test too.
RE: TESTING OF NUMERICAL RELAYS
All relay mnf. today not provide any data about test of relays. You can see: free maintenance!!!! 5-10 years warranty!!!!. Sales slogans only. You must test relay.
I don't want say about advantages of digital/numerical relays, its clear. new technolgies open for us new option and possibilies w/o end. But don't forgot with new technologies also come new problems.
1. Serial faults, you can see this type of fault only after some time, possible after 6-7 years. Found it at test time: NO CHANCE.
2. SW problems, same as point 1.
3. Multifunction relay with tens protection functions
you put on the first step only 3-5 functions and after 5 years you add new functions, you must, must check whole relay again as at commissioning time. Why, you can see some problems in SW, like to tasks problems, longer time of trip, CPU overloaded, event overflow, etc.
4. I don't say about algor. problems, we found this type of problems also and not once.
Don't forgot, new relay is math. and possible mistakes.
You can check relay and after few years found that relay operated with 25-30% error. Actually it's another problem, what relays you bay" you get, what you pay for ".
RE: TESTING OF NUMERICAL RELAYS
I would like first to define what "TESTING "means.
To my opinion, I will check Wikipedia later to see if I am right, and my using poor words, testing means "to carry a limited amount of 'experiments' with the system to be able to judge that it performs according tthe expectations"
Than we have to define "expectations".
But I hope you agree on that: ".. a LIMITED amount of 'experiments'..." that will allow me to judge the system. It the tests/experiments are not passed, we can say that there is something wrong.
If they are passed, we can say with a reasonable hope, that it should be correct.
We should also discuss on WHO is responsible to define the tests. Somebody who knows how it has been designed and know how it should perform. Somebody who knows where to put the "test points", like in the past (and still maybe today) with circuitry in the analog TVs, you had some test points on the printed circuit, where you were supposed to put your oscilloscope and somewhere else your signal generator and verify the correct waveforms... the test points were decided by the TV manufacturer! It was part of the design!
To my opinion, relay manufacturers today do not do it. The responsibility has moved away from manufacturers, and somewhere it should be brought to them, formally.
RE: TESTING OF NUMERICAL RELAYS
There is a measuring equipment and than a decision making equipment.
Both are made qith a technology that need "trimming", adjustment.
Simply, there is need to turn some potentiometers, trimmers, until the relay detects the correct point.
The correct point must be given from a signal generator (a calibrator generically. For "us" it is called test set).
You inject 1A (and you trust the test set that it is really 1A) and you turn the potentiometer until the relay trips or starts or whatever. Suppose that with this calibration, the potentiometer is set at 45 degrees. Another relay with the same name and manufacturer, will probably be set to 46 degrees, and another one maybe to 44 degrees. This because of tolerances in the components of the relay itself.
This is the basic of commissioning test. And this is the basic motivation that gave birth to test sets, Doble, Freja, Isa, Omicron etc. There was a clear need to have such equipments and we knew what they were supposed to do.
Then we know that there is drift in components/materials, so after a while we have to come back to the same relay and check that "it didn't move". The thresholds are verified again, and if there is need to adjustment, it will be done.
And if teh relay doesn't work? I mean, it does not trip, or I can't adjust it to the correct value, or I'll do it, and after 5 minutes it is far away from the given set point etc..
Nobody asked to the engineer (commissioning and/or maintenance) to take an oscilloscope and enter in the test points (if anyway available) of the electronic cards of the relay to try to understand why it doesn't work. Nobody asked the engineer to rewrap the coil of an e-m relay, or measure the magnetic field in the gap to see that it is really according to the design.
No, if relay does not do it's job, it was just declared broken or malfunctioning..
Of course today the problem is really this one: to be able to understand what means "if the relay does not do it's job", as the relay is today a complex system.
I don't want to enter already in the numerical world. I would like you to think about what above and eventually complete it, so that we have a clear panorama of the problematic in the "previous world".
RE: TESTING OF NUMERICAL RELAYS
I only one year worked with EM relays.
We had procedure difineded by mnf. for test those relsys, was included injection current/voltage and calibration of mechanical parts( springs, contacts,coils, etc).95% of work was calibration of mechanical parts. We also had special tools for calibrate them. As I remeber, for one single phase overcurrent relay we spend hours. As student, I worked at one commissioning company and get test protection of MV single speed motor. I worked on this about 4-5 days. Only test and calibration of 6-8 EM relays was about 3days. OK, I haven't excpirience too. Only one relay I tested fast,
static timer with knobs. EM relays was possible replace on the site or at some local lab.
Static world, good world, simple relays and simple test
No mechanical parts, tested exactly as you wrote. Calibration of knobs. Wat was problems , knobs in all cases is mechanical parts and those relays influenced by RF, mobiles phone and and .. havent connection to PC.
Before about three years we tested some generator protection with static and first digital relays( also with knobs). About 50% of relays was defected and reason was vibration of cubicles. I don't say about several unwanted trips becouse mobile phones. But in static relays you had "test points" also definded by relay mnf. and was possible tested them with scope and also possible repair them: replace some resistor, capacitor or repair at some local lab.
What is a big disadvantage of those relays: you test them put in operation and after 5 min you don't know it work or not, in static relay you see only one green led of PS.
One classical case: we tested EM diff relays and found all of them defected, but we tested them directly from terminals not from test plugs and after few hours we found that all CT's was shorted on the CT's terminals.
RE: TESTING OF NUMERICAL RELAYS
Next problematic point with analog injection, alsways you need some good Fluke for check your test set. Relay setting is 1A for 1s, you inject and have negative result. What is a reason, relay, test-set or wrong connection? What is a negative result: relay not work or you get trip after 1.3sec. Why, you don't know and start check all again. More problematic if it's angle, yes you can calibrate with knobs, but you don't know it's right or now. For exampe: we tested one reverse power static relay, relay passed w/o any problem, after one or two month was really reverse power and relay not operated. Why, we come again, test again this relay according to mnf. , all fine. Of course we found reason, solution was very simple: retrofit of protection relays.
RE: TESTING OF NUMERICAL RELAYS
I start from the point that the test set IS calibrated, so that it does what it is supposed to do. Ok with the Fluke of course, just in case you believe the test set is doing wrong. Instead of the Fluke you night consider the use of another test set (I am preparing myself for the discussion in numerical world, where the Fluke is difficult to have
RE: TESTING OF NUMERICAL RELAYS
Today so rainy and windy...
But seems we think at the same direction
We will continue in the digital world and after this at the numerical
RE: TESTING OF NUMERICAL RELAYS
So far:
1) what's the meaning of testing an e/m or static relay.
2) what are we asking to relay manufacturer
3) what are we asking to the test set
4) which test do we perform and why
5) which decision we take if test is passed and/or failed
so that this point is clear
RE: TESTING OF NUMERICAL RELAYS
http
of course it's empty!
Let's try to fill-out the hole in this Forum.
RE: TESTING OF NUMERICAL RELAYS
We work hardly too...
RE: TESTING OF NUMERICAL RELAYS
1. We also take in account digital relays, becouse I'm always separate digital and numerical relay.
=======
For first commissioning I cannot find much difference when testing of settings. May be I am wrong. Testing of logics is another story.
3.Commissioning time, full test:
3.1 Analog inputs
========
First point I test (and first table in my test report) is measurements test. Usually at the begining I inject current and voltages from terminal block where cables from CT's and VT's come, checking this way panel wiring. Next point is to inject from test blocks, if any installed. It is good to check whether short circuit facilities of test block operate without breaking the circuit. Sounds obvious, but last year I found that our Freja 300 cannot react on very brief breaks, during the switching of test switch. We found that wrong type of selector switch have been installed (without maintaining of closed circuit during switching) when tested second time relays using Omicron.
3.2 Digital inputs ( include threshold voltage operation), outputs
==========
Operation of DI's - OK, but threshold voltage check is not necessary according to me - there are factory tests for this purpose.
3.3 Settings
==========
Some manufacturers pretend that if measurements check is OK no need for detailed checking of setting values. I had a case when they required from me just to make shot test of one forward and one reverse point of distance relay characteristic and to sign commissioning report. I refused and never got other order from this customer. Anyway I don't regret.
For me initial commissioning must include checking of setting thresholds for each function. For numerical relays I don't see much reason to test PU/DO values - they are always good and are not dependable on ageing of some spring, for example. Of course more complicated functions require special programs, some of them included in test set software, some should be created by myself.
It is matter of discussion should test to be completely repeated if only some values have been changed during the test. Theoretically nothing wrong would happen, but we always have a risk something have been disabled and forgotten to enable again. According to me such risk should be accepted, but detailed review of printouts of final setting files must be done before energizing.
3.4 Logics
========
Most interesting point! This feature of numerical relays still keeps relay protection engineers necessary!
Actually I see here the biggest difference between numerical relays and all previous technologies. We cannot rely now on fully type tested configurations. Even when using typical configuration (like these ABB offers, for example) some minor changes must be set to adapt the relay to particular project. And main question is: Who takes the responsibility!? No answer for me.
3.5 power supply ( min and max thresholds).
============
According to me we should rely on factory tests here.
------------------------
It may be like this in theory and practice, but in real life it is completely different.
The favourite sentence of my army sergeant
RE: TESTING OF NUMERICAL RELAYS
I start from max requerements, you reduce all to gold optimum.
You wrote exactly procedure according to what we test relays.Logic it's other story, you are right, need decide case by case. About settings, in 10 years what I work with digital/numerical relay, we and our customers only once
found relay with errors in IDMT curve.
But, we have one customer with relays from some companies (w/o names) we tested those relay each year( some special requeremnts, critical object) and only on this years all relays passed, usually we was need replace one or two digital relays. Before , I think, one year we tested protection relays of one very serious company, four sets of diesel generator protection. We check all relays and after compaire results we didn't believe, same results for all sets with difference maybe 0.5% . crazy and this after about 8 years of operation.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
RE: TESTING OF NUMERICAL RELAYS
That's another example:
what does testing mean and what you do if the test fails ????????
I think you are talking about the test switches of "Asea type", where you enter the test handle and it automatically shorts the CTs, cuts the VTs and trip signals.
There is a potential flashover in those equipments, during the insertion/removal of the handle. The flashover is generated by the CT itself, which is a inductive circuit, and is not dangerous. Actually the flashover makes that the current circuit is not interrupted. No danger.
I think you have just tested the test switch in a no load situation, and used Omicron to detect a small and fast contact opening, which mechanically may happen, but the circuitry where it is inserted makes it correct. They are designed like that. You could have used an oscilloscope to sense the switching, but it was not -if I well understood- a realistic situation.
So I think your "old Freja" did its job, by not detecting it.
RE: TESTING OF NUMERICAL RELAYS
521AB, I don't like to name the companies, but will make an exclusion now. It was Siemens design, but not the funny test sockets with selector switches which Siemens call "test switch". It was specific circuit, because primary design is double bus bar with transfer bus and it was necessary to switch CT's from the feeder and from the transfer bay to same relay protections. Design was OK, but in the factory they didn't order properly selector switches. Just mistake in delivery, it is good we caught it before energizing. Possibly operators never will switch them on-load, but I hate to miss such things anyway.
By the way I like these Asea/ABB test blocks and still keep test handles for both modifications. But I prefer switchable terminal blocks, test blocks according to me are just one more break point in a very sensitive circuit.
------------------------
It may be like this in theory and practice, but in real life it is completely different.
The favourite sentence of my army sergeant
RE: TESTING OF NUMERICAL RELAYS
See attached photo from our first cubicle where used only
URTK/S. up to terminal 12 are currents terminals.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
"...Design was OK, but in the factory they didn't order properly selector switches. Just mistake in delivery, it is good we caught it before energizing. Possibly operators never will switch them on-load, but I hate to miss such things anyway...."
you mean that the current circuits were OPENED by the test switch? And it was not a short interruption? I wonder why Freja was not able to detect this.
Regarding companies, they do exist, and as long as we intentionally damage / favour them, I think it is not a problem in naming them.
Omicro, Freja, Doble are also companies, and I see that nobody has troubles in naming them.
RE: TESTING OF NUMERICAL RELAYS
Strange for me but Freja didn't react on so fast breaking of current circuits. Omicron's software has indicators on the right side of the screen which turn red even after short interruption.
------------------------
It may be like this in theory and practice, but in real life it is completely different.
The favourite sentence of my army sergeant
RE: TESTING OF NUMERICAL RELAYS
I was working at that time with the "ASEA" test switches (I say ASEA because it comes from the old and good Combiflex time), that are designed to switch between main CT and test set while the substation is in operation, so under primary load.
RE: TESTING OF NUMERICAL RELAYS
Slava
RE: TESTING OF NUMERICAL RELAYS
I think mention should also specifically be made of Scheme Tests vs Relay tests. I quote seperately for a customer for these two items as a scheme tests involves the relay tests AND checking scheme circuitry - e.g. operation of switches, SCADA I/O, cb trip/close tests, etc.
With relay testing I am now referring to commissioning a new relay. I hardly do maintenance testing. I always test all protection functions, I/O and logics. I write my own macros for our test set and I verify results against manufacturer tolerances.
a) OC/EF relays - relay pickup and dropout and check three points on curve
b) Directional Relays - prove the zero torque line
c) Definite time - relay pickup and two points for timing test viz. at 2x and 5x pickup
4) Differential relays - verify minimum diff setting, slope 1, turning points, slope 2, diff hi-set, 2nd and 5th harmonic. Note that I actually test the EXACT throughfault stability characteristic of the relay and not just one shot inzone and one shot out of zone.
5) Distance relays - X reach, R reach, halfway up the right blinder and halway along the top line (quad). Need to modify depending on whether load encroachment used.
I also do the above for our local utility. I think it extremely important that an engineer who FULLY understands how the relay thinks, compile the test sheets. Modern relays have some very involved algorithms and the mathematics can be quite daunting. Just consider how modern digital distance relays do EF compensation for example or how to calculate test points for a current diff relay.
We thus have a policy here (at the utility) that the engineer accompany the technician for commission testing. Maintenance testing is performed by a technician and is more along the line of spot checks as mentioned above. For my industrial clients I prefer to do all commissioning myself. I do not believe it to be in the best interest of the client to subcontract the commissioning of a protection system when I was involved in the design and settings.
I am sorry to say but you can forget about getting expert technical support from your local supplier (rare if you do) - in this country at least. Most times when I analysie the relay my queries get routed to the designers in another country. In my experience many manufacturers do not have the indepth technical expertise and experience locally available to assist with detailed technical queries. The local agent is more a front end customer service rep. In the US it may be different I do not know. Very often they may be able to explain a setting but cannot give recommendations as to how the relay should be set as their power system knowledge/experience is rather limited.
There are exceptions I must add and I am very grateful for their assistance.
Hope this helps.
Regards.
RE: TESTING OF NUMERICAL RELAYS
It's nice see,how we all think at the same direction.
I'm so happy. Not only with test procedure, also work with utilities and industrial clients.
For Plamen: If you write logic or change mnf. logic,original setting: it's your full responsible, same for protection, same for control, same for settings. Yes, I know, you not responsible for setting calculation, it's responsible of clients, but what you do, if you see mistakes. I always change it, but it's not good from one side, from other side I would like sleep at night.
This is another topic, where is a limit of our responsible on the site.
About local agents: yes, it's not so simple, usually are salesmans, but it's depend on country and companies.
Now, about why responsible for the test procedure.
Mnf, I don't think so. Commissioning eng of this mnf., yes, in lot of cases, but not always.
Client, if are utilities, yes 100%, only them.
Industrial clients, this is problem of problems:
1. Client, in lot of cases, havent any excpirience in protection,
2. Designer??????????? only if it's control function and simple protective functions.
3. We????designer and commissioning eng. I'm not sure, if it's right. We do this and always, but isn't correct. Need some another person or company.
I think about it, but haven't any solution.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
It's great, good, etc, etc.. Omicron, Frej, Doble ( with additional amplifiers) and other companies too.Great test sets, no words.
Only one not technical problems, new commissioning guys work with this Macros, according to Plamen's or Verita's or 521AB's test procedures and don't understand what them do!!
For this, before one year, we bouth some test set like to old Sverker 650,w/o communication and I'm request test all simple rely with this test set only, BTW I'm used it always.
RE: TESTING OF NUMERICAL RELAYS
"... About new technologies in the test sets.
It's great, good, etc, etc.. Omicron, Frej, Doble ( with additional amplifiers) and other companies too.Great test sets, no words..............................For this, before one year, we bouth some test set like to old Sverker 650,w/o communication and I'm request test all simple rely with this test set only, BTW I'm used it always...:"
I must say I understand you. Problem is that test sets took their own way, without feedback/contacts with relay manufacturers, and this is not only their fault, it is ALSO their fault. Even relay manufacturers have their responsibility in this missed dialog, so the risk is today that you test one relay from company ABC by using test set and procedure XYZ and you ... have not tested the relay at all, or you have tested it wrong.
At the end, you get one SVerker.. I see the point and I enjoy this forum
RE: TESTING OF NUMERICAL RELAYS
It is our National Day today, so I stay a bit outside of professional problems. Tomorrow may be (I have to complete some urgent order) I will be again on-line in this discussion.
Slavaq, BTW I have in my practice three distance protections tested only with Sverker 760. It was in Lagos and there was nothing more suitable around. It is possible to do the work this way, but I wouldn't like to repeat it again
------------------------
It may be like this in theory and practice, but in real life it is completely different.
The favourite sentence of my army sergeant
RE: TESTING OF NUMERICAL RELAYS
Sverker 760...., isn't so intresting
521AB, btw, Combiflex is still goodest aux relays, when possible, I'm used only them, 30-40 years of product, like to ASEA RI relay, 67 years of product.
RE: TESTING OF NUMERICAL RELAYS
Understand me right
Good day to you all.
Slava
RE: TESTING OF NUMERICAL RELAYS
And I think there is a lot of true in it. Test sets are getting complicated, and are starting to give non-realistic problems to customers and, as consequence, to relay manufacturer.
There should be a sort of high level decision (IEEE, IEC, ANSI? I don't know) to decide -at least the minimum requirements- on what should be test and how.
It should also be decided what is a commissioning test, what is a performance test. Today people test with steady state shots and believe to find "bugs" in the relays, while actually the relays find bugs in the test sets.
THIS was what originated my thread.
I don't thing we will manage because to my opinion, nobody has the answers to this. But let's see..
RE: TESTING OF NUMERICAL RELAYS
Its remind me some test with customer:
We tested pole slip generator protection with new Doble, test was about one hour,50min was calibrations of test set and we check this test set with this generator protection
We will continue. and will manage it
99% it was your mistake.
I also would like say again, if I bay good relay and we bay only good relays, I check only meas, BIO , tripping and signalization logic, its all, not setting, not drop-out, nothing, no needed. But first time we check new type of relay tens time. Im not KEMA and not type test company, Im only commissioning eng. Sorry for metaphor, if I bay and believe to good car, what is a difference, need believe to relay mnf. too. I work only with two relays mnf. and believe to them, of course check and test relays, but not all parameters according to manual. Guys, today, with new concepts, double protection, back-up of back-up, chance of not operated about zero, unwanted trip is possible,so, is not good, but I have statistics for ten years of my work.
NO NOT TRIPPING CASES. ONLY TWO REALLY BFP OPERATIONS. ABOUT 10-15 UNWANTED TRIPS ( wrong setting or HW problems). About 10 UNWANTED trips becouse relays fault, its all. For my pinion is nothing. I told about 50 substations and about 6000 installations.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
1) "Im not KEMA and not type test company, Im only commissioning eng."
The same for test sets. What are we asking them? Commissioning/maintenance test. NOT TYPE test tests, NOT relay performance tests.
Or let me say better: test sets can do smarter things, like performance tests and/or type tests, but they should declare what they are for, and put all the tests together, and move all the responsibility to the test guy.
If we decide what a commissioning test is, we have done a good step, and I think you have some good ideas, otherwise you didn't write "I am not KEMA...".. Just try to formalize them!
2) "...if I bay and believe to good car, what is a difference, need believe to relay mnf. too."
Exactly! I was discussing this with some friends of mine (and colleagues).
When you buy a new car, do you start to test the ABS functionality? Or you ask for the report of the Otto cycle of the engine?
No, but on the other side, you perform some tests, even if not formally.
3) ".... NO NOT TRIPPING CASES. ONLY TWO REALLY BFP OPERATIONS. ABOUT 10-15 UNWANTED TRIPS ( wrong setting or HW problems). About 10 UNWANTED trips becouse relays fault, its all. For my pinion is nothing. I told about 50 substations and about 6000 installations...."
This is an interesting statistics. What voltage level? Do you have a main voltage level or you "run everywhere?"
RE: TESTING OF NUMERICAL RELAYS
RE: TESTING OF NUMERICAL RELAYS
Pink Star near to your name: THANK YOU FOR YOUR POST!!!!.
My statistic is include voltage level from 3.3kV up to 400kV. But lot of relay are MV level and generator and transformer protection.
About commissioning. I would like ask:
We bay not satnd-alone relay, we bay some BBP.BFP system for
28 bays with tripple buses with 12 protection zones and additional aux. bus, with couplers are included six disconnectors in distribution version. Folks its example only. What I can do, check each unit as stand alone relay, connected 20 synch. test-sets for check stability and performence of relays. Check again all zone configuration.
Of course no. FAT of this system is maybe two days. That means I must believe to relay mnf and to test eng. On the site, I provide only small testing for check AI, BIO and statuses of HW eqp. Its all. And of course primary stability test. Now, what is a difference with alone OC relay, no difference. In many cases no needed any smart test sets on the site. Simple Sverker enough also for distance, pole slip, loss of excitation, NPS, AR and diff protection!!!!!. Problem is only function like to ROCOF, freq, etc.
I always say to my bosses, customers and to my team:
Commissioning of relay on the site is connection of this relay to the system (CB,CT,VT, Alarm Panels, SCADA). NOT PERFORMANCE TEST OF RELAY.
Regards.
Slava
RE: TESTING OF NUMERICAL RELAYS
Of course NO!!!.
RE: TESTING OF NUMERICAL RELAYS
I'm in the USA. When I started in this business, everything was electromechanical. If you needed overcurrent, you had three overcurent relays (or maybe four). If you needed phase current balance, that was another relay. Undervoltage? Another relay. A large motor or transformer had aalmost a dozen different relays, all made with clever configurations of iron and copper, and when it was time to test them, you dutifully pulled each relay out, did careful inspection, cleaning, contact burnishing. You knew the realy worked per the manufacturer's manual.
If you wanted to see how the relay functioned int he circuit, you rolled out the drawings and saw how the input currents and voltages came in, and where contact action would affect the circuit, and it was ll right there on the drawing. Life was good.
Fast-forward to today. Instead of that dozen relays protecting my transformer, I have ONE GE UR T-60. It has all those old functions: primary and secondary overcurrents, neutral, ground overcurrents, much more ocnfigurable differential and restraint characteristics. The siring is much simpler: Two sets of currents and a set of potentials in, contacts out.
But information that used to be on the drawings as to how the different protection elements interacted int eh control circuit, that information is not on the drawing as it was. It's in the relay's programming. Where it used to take a real effort to add wiring or cut connections to change the function of relays in a control circuit, now it's a matter of an ambition, if mis-informed engineer or technician with a laptop computer to foul things up. I've seen that done.
I've also seen setpoint programming that was incorrect, i.e., essential elements not included in the trip logic statement, or the incorrect implementation of the elements included resulting in misoperations.
All this means that our job got both easier, install one box and we have just about every function a protection engineer could dream of, and it got harder, we had to make sure we did our work with a manual the size of a metropolitan city's phone directory, and anybody with a laptop could mess things up without any external evidence of future problems.
And all that takes us back to the original post: What do we do about periodic maintenance testing of digital relays?
I am of the opinion that the most important things are to verify that the relay is still properly processing inputs, in other words, it properly reads voltage and current inputs, that the outputs still react properly, i.e., a contact closure results in the proper action the control circuit, and finally, that there have been NO changes in the programming of the relay since the last intentional changes.
Note that I said "periodic maintenance testing". Obviously, commissioning and type testing require a much different approach.
old field guy
RE: TESTING OF NUMERICAL RELAYS
One US guy with us, one Australian and three from Europa. Not bad statistic.
We still wait to other.
I think, we dont need discuss about type test. Only commissioning and maintenace.
RE: TESTING OF NUMERICAL RELAYS
We are a pretty good cross-section of the electrical power business, aren't we?
old field guy
RE: TESTING OF NUMERICAL RELAYS
Thanks oldfieldguy for coming in.
Regarding the star: is it possible that as soon as I write a post, like this now, somebody clicks on "Thanks 521AB for this valuable post"?.
There must be a bug somewhere!
RE: TESTING OF NUMERICAL RELAYS
I just sent that bloody "Protection coordination study" which ate-up all my time and attention during the last week. So now I can add some new 50 cents on the topic.
Slava, I am not 100% agree with you about threshold testing. I believe they should be checked regardless of measurement test of the relays. Not that I don't trust type-tested algorithms which produce relay characteristics, but mostly to check how correctly setting values are entered. It is especially valid for Thermal Overload function and IDMT characteristics, where it is not so easy from the first look to determine tripping time for certain current.
Veritas specify almost the same amount of tests we do with one exclusion: for differential relays we do only threshold test, not slopes. The first reason is that we haven't equipment (6 current channels) for slope testing, but actually nobody have asked for this up to now. According to me it is not absolutely necessary, I rely on relay algorithms for that.May be I am wrong, but I prefer to keep the minimum necessary tests which are required by the customer and which also allow me to sleep well. If/when we can afford more expensive test set probably we will do slopes also, even to make test more "scientifically looking", as my professor in the university said. With automated procedure it will not take longer time and definitely will reduce understanding of my guys on the amplitude and angle compensation - now it is a bit tricky to determine expected operation values depending of vector group and CT-ratios!
I like most of comments of the Oldfieldguy, some of them (about manual the size of a metropolitan city's phone directory ) I often say with small language variations
Could you give me some information regarding the legal side of relay testing in different countries. They put here in Bulgaria some requirement to establish in the company special Accredited Control Body for most of tests concerning safety - like insulation resistance, HV tests, earthing resistance, electrical installations, etc. And sometimes it is required that relay protections also to be tested only by accredited Control Body.
Problem is that accreditation is completely bureaucratic procedure, something like certification under ISO 9001, but much harder. According to me it is established only to take money from really working people, never mind how well it looks on paper. I am curious how this is arranged in other countries.
OK, have a nice day/evening depending of your location on the globe !
------------------------
It may be like this in theory and practice, but in real life it is completely different.
The favourite sentence of my army sergeant
RE: TESTING OF NUMERICAL RELAYS
I wrote some extrimaly points and you correct me and correct right. I would like add, you must test all IDMT functions and "thermal" function like to: V/Hz, NPS curves, Overload. As Veritas wrote: 3 typical points:
at the begining, middle and end of setting. Its very important point, IDMT curves have differnt errors depending on the fault currents from 12% up to 3%.
And, I would like back to the begining of thread, I wrote:
at the new numerical relays important CPU load and right tasks and modules, for test this, you must provide some "type test" of your relay configuration and settings and correct SW or file.
Regards.
Slava