You are on page 1of 2

Behaviour of Protection relays under CT Saturation

We all know that ALF (Accuracy limit factor) determines the maximum limit of CT at which accuracy is maintained and it is expressed as a multiple of CT rated primary current. Now consider following scenario: A transformer 11kV / 433 is connected straight to a grid and a CT is provided on HV side for O/C & E/F protection application. Grid fault level >20 kA HV side CT details :5P20 at rated burden, ratio 100 /5 Transformer HV full load current - 100 Amps, % Impdance of Transformer 5 Maxm through fault 100 /5% = 2000 Amps Known: With ALF 20 and rated primary current 100A, CT will maintain 5 % accuracy up to 20 times of 100 Amps i.e. Up to 2000A. Hence correct/accurate operation is ensured for maximum through fault level of transformer. We also know that Knee point voltage is directly proportion to If and ALF Now comes question . At grid fault level (>20 kA) what are the chances for O/C & E/F protection to operate as CT is already saturated. Since grid fault level (>20 kA) is more than the designed ALF factor of CT (for this case it s just 2 kA).Do we need a bigger CT with high ALF close to grid fault level to address the issue ? Is there any reference where grid fault level is used to determine ALF for better accuracy and to ensure protection operation? And the Answer is ..NO Please read through for explanation.

When we say ALF is 20 it means that the CT guarantees that it will transform up to 20 times the rated current within the claimed composite error with the rated burden connected. This means if the CT is 500/5, 5P20, 20VA - it will transform up to 10,000 A of maximum fault current when the secondary connected burden is 20VA. However if the secondary connected burden is lesser then it will transform higher currents (the effective ALF increases). This is true with modern relays as mostly the burdens are very less and so the effective ALF available is quite high. Secondly even if the CT saturates, the secondary current is not completely zero, it s distorted. Now depending on the degree of saturation, type of relay measurement (peak, Fundamental, true rms, etc), the relay may measure currents much higher than the setting value even with saturated waveform. Thus the relay in most cases will operate. If it was an IDMT relay the time of operation may become higher than expected ( as the secondary rms or fundamental current will be lesser).

Now for some applications where the grid fault current is very much higher than the CT rating and ALF and the connected burden is close to the rated value, there has been a practice to provide CT with higher ratio so that it will transform the full fault current. One typical case would be the short circuit protection of motors. The bus fault level will be controlled by the feed to the bus and can be very high. If a low rating motor is connected to that bus, the load current will be less and so the CT ratio selected will be very less. All protection for that motor (overload, NPS, etc) will be connected to that CT. However only for short circuit protection a separate CT with higher ratio used to be provided to ensure proper operation (as instantaneous setting itself will be higher). Now this was more prevalent in electromechanical times than in Numerical relays as the relay burden compensates for this problem in most cases!! Further to above Providing Vk of "If ( 1+X/R)*( Rct + 2 Rl)" ensures that CT does not saturate for the worst situation. But even this does not account for remanent flux. The amount of remanent flux can vary very widely. It would not be practical to consider a specific value of remanent flux, thus its not explicitly considered. However since the probability of maximum fault current, considered X/R and point on wave of fault which gives maximum offset are very rare, we can say that the above value does have some inherent factor for remanent flux in many situations. However still we hear once in a while cases of undesired performance. Now depending on the type of operating characteristics and the measurement algorithm (if its numerical), the relays can accommodate certain amount of errors. This can be brought out only by testing the relay (or its model) using real time systems. This is what is called "Conjunctive" tests. Earlier this used to be done using real CTs and relays with fault currents of real magnitudes. Nowadays with real time digital test systems and softwares we can do it by simulation itself.

Conjunctive tests are done for all the Numerical relays to arrive at the correct Dimensioning factor of CT. Now a days with RTDS we can actually arrive at the minimum voltage required for the relay to perform within the guaranteed reach accuracy, and guaranteed operating time. In RTDS we simulate DC transients , Remanent Flux ,Build up of flux due to repeated reclosures and then testing is done for reach accuracy and operating time. In the test we could arrive that Protective Relays can work even under partial saturation. But if the CT is designed with KPV less than the value asked for , then for a fault with CT saturation , the relay will maloperate ( reach accuracy problem) or operate with a time delay ( DC transient will die down after a period of time depending on system X/R and subsequent to that CT may come out of saturation ) or non operate. Now for the catch: All the CT requirements given by different manufacturers have a validity condition attached. Some explicitly mention these conditions, some don't. I remember we (in AREVA) used to give the maximum through fault current and X/R ratio for which a particular Vk formula is valid. This was true even in our static relays. I guess this practice is followed even in numerical today also. So if you are comparing Vk requirements of different relays, its necessary to understand what are the validity conditions proposed by each manufacture. As the proposed Vk may only ensure through fault stability for 100ms.So if the external faults are not cleared fast, then the relay may loose stability after 100ms. The moral of the story is that don't get fooled by somebody claiming they require very low Vk unless it s very clearly explained under what conditions.

You might also like