Sunday, September 28, 2008

The Physics of Failure of Transformer


Last time I promised to answer Mr. Chattopadhayay's queries on the blog "Funny and Freak failure mode of a transformer".

First of all I would like to start the explanation with my most favorite style -- i.e. by understanding the Physics of failure of a transformer and then explain the physics of how the explosion would occur and what its consequences might be. I would then try to highlight other possible reasons of such explosions taking place and conclude by listing out the possible measures that can be taken by us to prevent such life threatening explosions.

Any short circuit of such a transformer would produce an electric arc that contains very high heat energy. This high energy is then capable of initiating a phenomenon called 'oil cracking' (this principle I believe, is also used favorably in petrochemical industries). What happens when the transformer oil cracks? It produces primarily acetylene and hydrogen as dangerous combustible gases. However, with high di-electric strength of the oil along with the inherent good design of a transformer it is difficult, if not impossible, to produce the cracking phenomenon. This is because all high voltage power transformers are to be tested for short circuit after these are manufactured. This is done as per IEE norms of short circuiting. The transformer is short circuited and kept that way for at least 2 sec, which is sufficient for the cracking to take place, since such cracking would take around 0.1 sec time to happen. Unfortunately very few transformer companies have such testing facilities and very few users specify such a test during procurement stage of a power transformer.

But coming back to the point, these combustible gases can now exist in at least two ways -- either as oil in vapor phase or as mist having explosive properties. The high energy of the short circuit then easily sets off the fire to the combustible gases, which then rapidly expands and explodes producing rapidly oscillating pressure waves. And as we know, the pressure of such waves is directly proportional to square of acceleration. So, it is the acceleration component that produces the high force that explodes the transformer spilling the oil and spreading fire all over thus damaging property and possibly human lives.

But what can produce this short circuit? Well the unwitting hawk could have certainly caused it. But there appears to be another common reason. Such short circuiting might also happen if the bus bar support barrier catches moisture from the atmosphere when we open up the transformer for repair and maintenance (especially in places having high relative humidity). This, I presume would generally go unnoticed by the maintenance crew if they are not very enlightened about the physics of failure. The moisture then compromises with the insulating property of the barrier and is subjected to electrical tracking that leads to a definite possibility of short circuit and consequent fire and general all round failure. In that case the cover of the link box explodes first. The failure analyst can look for such important clues at site to establish the cause of failure.

But what can we do to prevent such explosions and loss?

There are a number of possibilities some of which are as follows:

a) Specify short circuiting test during procurement stage (the best option)

b) Install Transformer protector. This is a passive mechanical system that is activated by the level of transformer tank internal pressure reached during a short circuit. At least two transformers can be simultaneously protected by one device.

c) Install protective system based on the release of superheated water at 180 degrees C.

d) Install protective systems based on the release of powder to suppress the fire and subsequent explosion.

e) Install protective systems based on water and salts.

f) Regularly monitor the condition of the existing protective devices and the general health of the transformer.

It might be of some interest to see the photograph of the burning transformer and the unfortunate dead bird, which I have pasted on this blog.

Hope this would be of help to Project Managers, Plant Heads, Safety Managers, Maintenance Managers, Electrical Engineers and Reliability Specialists to prevent such unfortunate events in our plants.

Signing off for now,

Dibyendu

Saturday, September 27, 2008

Making the Most of Customer Complaints, - MIT Sloan Management Review

Making the Most of Customer Complaints, - MIT Sloan Management Review

Service Failures (more commonly known as Customer Complaints) are also failures like equipment failures that need the careful attention of managers.

This article provides interesting insights into dealing with service failures. It means a lot more than just fixing the immediate problem, which we usually do to address any type of failure.


The solution proposed in this article is to address tensions that arise among front-line employees who handle complaints, the managers of those employees, and the customers themselves. Steps include starting a complaints database that would induce managers to think out of the box to improve service, and rewarding service employees not for reductions in complaints but for providing exceptional solutions to problems.

But what is intuitively not clear is that all product complaints are always linked to the design of our factories, the design of our machines and its upkeep and the design of the associated processes. I have seen only a few companies that make a direct link of complaints to plant engineering (I am not somehow comfortable with the terms like mechanical maintenance, electrical maintenance etc since it puts people into narrow silos and kills the joy of enjoying the engineering job in its totality). Hindustan Unilever, as I have seen in one of their factories, directly addresses product complaints through shop-floor improvements and redesign of their machines, components, materials, packaging and processes. They take up this activity very seriously and often succeed in solving problems permanently through innovative methods thus bringing about a change in the number of product complaints. Tata Metaliks is another company I have seen in my fairly long career that also tries to address customer product complaints through innovative improvements in the process and equipment. However, improvement in a process industry is far more difficult than making improvements in a FMCG industry. This is because process industries are far more capital intensive and once a machine is selected and installed it is very difficult to undo the mistakes done or modify the system.

But unfortunately, actively linking product complaints to shop floor improvement is not the case with most companies I know of. Why is this? This is because, companies that succeed are companies that are managed by people and managers who prefer thinking out of the box and concentrate on the quality rather than on numbers. However, the most important reason, as I see it is that these organizations are determined to succeed. And such people or cultures are difficult to get or build. The intention and the will of successful organizations are clear. And such organizations are plain lucky to have such managers and employees on their roll. They build the much needed courage amongst others in the organizations to play, experiment and fully enjoy their jobs. Because innovation is simply 'serious play'.

The worst case, I have seen over the last thirty years is that of Hindustan Motors. They just refused to improve their machines, design, materials and processes. They could have done fantastic innovations through simple engineering improvements. But they stubbornly refused to listen. The result is that they have now almost managed to push their product (their good old Ambassador cars) and themselves to oblivion. Such companies behave like 'ostriches' that refuse to see and acknowledge the problem as it is and do strange things believing something else would bring about their profitability. Hence they simply lack the system to improve. They either believe that 'quick fixes' and 'clever manipulations' are enough to do the job or simply don't believe that 'customer pays for their product'.

So, are you linking customer complaints to engineering improvements in the shop-floor or the factory?

To read the main article just click on the link.

Signing off
dibyendu

Sunday, September 21, 2008

Response to 'Freak Failure Mode of Transformer.

I must thank Mr. Partha Chattopadhyay, Chief Engineering of Tata Metaliks, Goa plant for his kind comments in the form of an email in response to my earlier blog on ‘Freak Failure Mode of Transformer’. I value his comments since it comes from a fine electrical engineer know of. He brings in his original perspective on the issue? But I have a more important reason for putting it up in my blog since I have grown up as an engineer to believe that any engineer can learn more through such interactions of live case studies and increase his/her width and depth of knowledge which would ultimately benefit our society in many ways.

However, as desired by him, I assure him that I shall try to put in my thoughts on the issue of fire and explosion of transformers more cogently in my next blog entry in greater depth and try my best to explain the physics of this phenomenon. Fire and explosion of transformer is a serious issue which is often neglected or not properly understood even by experts on safety. However, once understood properly, plant engineers can take care of this serious issue during the procurement stage and during operation and running of the plant. It is also important since it affects so many people and society at large. And it might be of interest not only to electrical engineers but also safety engineers, project managers and other plant engineers too.

Signing off for now

Dibyendu De

Quote


Dear Mr. De,

I have following points for your consideration:

1) Transformer oil’s flash point is around 140deg Centigrade. High Dielectric strength only ensures that no spark occurs (which means breaking of insulation and passage of huge current, which may lead to attainment of flash point) between to points having high voltage gradient. Transformer oil having good dielectric strength will catch fire easily provided it attains its flash point.

2) Also, DGA helps us in finding incipient fault of transformer winding. Generally, cotton and cellulose materials are used for transformer winding. Traces (ng/L)of Methane, Ethane like gasses can be detected in the transformer oil. Such traces can not contribute further to transformer oil’s natural inflammability.

I guess that a huge spark might have occurred when the hawk touched with the terminals of the transformer. Such short circuit will certainly generate massive electromagnetic force, which can cause mechanical damage to the insulators, resulting into leakage of the transformer oil. In addition, I feel there was an accumulation of transformer oil on the top cover of the transformer due to some unattended leakage. Probably this accumulated oil caught fire first and then the leakage contributed to continuation of fire.

Following measures may act as antidotes to this type of trouble

1) we must not neglect small issues like checking of all safety devices 2) periodic oil testing, 3) keeping the transformer free from leakages, 4) periodic checking of transformer’s protective devices, 5) Use of fire retardant synthetic oil(Very expensive!!).

Yes we need to institutionalize maintenance audit. This will certainly help us in finding systemic gaps.

Looking forward your reply,

With regards,

Partha Chattopadhyay

Unquote


Dibyendu De



Thursday, September 18, 2008

Funny and Remote Failure Mode of Transformer

On 16th Sept 2008 (Tuesday) a high voltage distribution transformer caught fire in Behrampore town in the state of Orissa in India plunging large areas into darkness for two days (Times of India, dated 17th Sept 2008, Kolkata edition). And how did it fail and what was the reason?

Apparently a hawk flying over the sub-station hit the high voltage electric wire and fell on the 31.5 MVA (Mega Volt Ampere) transformer which instantly burst into flames. Evidently, due to a mechanical fault, the auto-fire protection mechanism of the transformer failed to work and when departmental staff tried to douse the flames with water, the flames rose higher and went out of control. It is interesting to note that the transformer contains 20, 000 liters of oil to keep it cool and act as insulator.

Apparently it seems that it was freak failure mode. But a little amount of careful thinking would show that it is not. The problem reveals a lot of issues and imperfections in the system.

First, why was the mechanical fault of the auto fire protection lying unattended? Simply because no one perhaps monitored whether the protection system is functioning or not. This is called detective maintenance.

Second, why did the departmental staff try to put out the fire with water? Can an electrical fire by put off with water? It means they did the action without any awareness. They simply thought that it is an ordinary case of oil on fire. Even fire on oil should be doused with foam and not water. A gross neglect of training and lack of preparedness.

Third, what about the strength of the transformer? Was it tested for short circuit resistance? It is not supposed to breakdown if the transformer is tested after manufacture. What about analyzing the gas build up within the transformer? What about regular venting off the gas? Was it built into the system? What about regular and simple monitoring of the di-electric strength of the oil and the dissolved gases in the oil? What does it show? No care to monitor the condition of the transformer and no regard to the intial specification of the transformer during purchase.

So, we can conclude that the event not as an freak accident but as a gross neglect of maintenance and reliability thinking that led to the loss of equipment, manpower and loss of power for the consumers. Even if the striking of the hawk appears to be remote, it is not the real cause of the failure. The real cause was human neglect to care for their assets that led to the failure. Reliability and Maintenance pays!

Dibyendu De

Wednesday, September 17, 2008

Lip Seals -- Failure rate Monitoring and Energy loss

Some manufacturers of lip seal did a research work on the life of lip seals and they found that on the average the Mean Time to Failure of lip seals was around 77 days of operation or around 1644 operating hours. And the maximum life is around 3000 hours or around 180 days of operation.

But there is another disturbing fact. The same group of researchers also found out that the energy loss per lip seal is around 147 Watts of power. So, it is not difficult to calculate the total loss of energy when we are using more than 500 lip seals all around the plant for 24 hours of operation. It is a substantial amount of money.

So, I feel that there are three important issues to the question of reliability of the seals.

The first issue is how do we improve the life or durability of the seal?

The next issue is how do we monitor the condition of the seal? There seems to be no proven method of monitoring the condition of the lip seal except for the fact that some engineers have tried out the use of Infra-red thermography heat pictures to monitor the condition of lip seals, which I think is neither too elegant not does not seem to always work.

The third important issue is how do we reduce or contain the energy loss due to use of lip seals?

Usually we understand the defect after the failure has taken place. And the consequences of the failure can be high, including safety like the contained fluid catching fire. Evidence of secondary failures like grooving in the shaft is also not very uncommon to find.

Hence this is a critical failure mode that needs to be addressed in a variety of ways, like proper selection of shaft material, hardness of the material, direction of grinding of the bearing and seal areas, surface roughness, lowering the coefficient of friction, minimizing the effect of unbalanced forces in the system and alternative use of non-contact seals. We have tried out these measures improve durability of the lip seals and it works.

Since a direct monitoring method is yet found the alternative is to use a probabilitic way of assessing the life and condition of the seals by measuring parametes that has a direct influence on seal life. This is possible and I found my method to work in all cases and conditions. The only problem is it is sometimes difficult to establish the 'probabilistic parameters', which may vary from case to case.

However, reducing energy consumption is again a tricky issue. This can also be dealt in a probabilitic manner and by keeping the influencing parameters within control or to an optimum limit of a probabilistic combination.

Dibyendu De
Design Innovator -- Plant Reliability

Tuesday, September 16, 2008

Heat Flow and Reduction of Power loss

Air Conditioning Power Loss and its Prevention

As more and more air conditioners enter our homes in India we have a problem. Now everyone starts thinking of their power bills. And the utility companies also start worrying about the sudden increase in loads and dynamics of the situation. As we have seen that demand outstrips the supply position. And this is one of the reasons of power cuts in our cities and villages, especially during the summer months.

So, what can we do about it?

As consumers we can think of reducing the power consumption of the ACs. We can best do this by thinking about the direction of the heat flow from our rooms to the ambient. Since the ambient is at a higher temperature than our rooms heat would always like to flow from outside to the inside. And what do you think is the most likely passage of the heat flow?

Though we seldom think about it the maximum heat flow takes place through the glass windows. And the heat flow is through radiation. Hence, all we need to do is to paste a film onto the outside of the glass windows to reflect the radiated heat and prevent it from entering the room thus reducing the load on the air conditioner.

What the electricity utility companies can do to meet the erratic surges in demand? They can along with their coal based thermal systems have a number of gas turbines which they can press into service as the demand goes up. This is simply because of the fact that they can store gas whilst they can't store electricity.

Dibyendu De
Design Innovator -- Plant Reliability

Saturday, September 13, 2008

A Note on Failure modes of 304 SS and its prevention.

Chromium in concentrations above 12% renders steel stainless. 304 stainless steel, which contains 19% Cr and 9% Ni. (The nickel strengthens the alloy and gives greater corrosion resistance.) Unfortunately if the steel contains nearly 0.1% residual carbon from the iron and steel-making processes then the chromium in the SS has a strong affinity for carbon, and slow cooling through the red heat range allows chromium carbides to nucleate heterogeneously on the grain boundaries. The adjacent regions in the grains are depleted of chromium to far below the 12% threshold and are no longer as corrosion resistant. Thus, the steel is said to be "sensitized" and is susceptible to intergranular corrosive attack. The slow cool after the welding of the 304 SS allows such precipitation and triggers sensitization. This is technically known as the sensitization temperature of SS which is around 650 degrees C (ball park figure). It can get activated during welding or due the application like using 304 SS in very hot working conditions.

Even when sensitized, the steel is adequate for many applications, such as household products (SS utensils that we use on the gas ovens) and even containers for less concentrated nitric acid. However, while sensitized steel is adequate for 75% nitric acid, it could not be used for the 90% solution.

Whenever corrosive attack on 304 grade of SS is the predominant failure mode it might be prevented in several ways. For example, low carbon stainless steel, designated as 304L could be used. Or, addition of niobium during steel making would tie up the carbon as fine, harmless intragranular niobium carbides. Alternatively, we can anneal, especially welded joints on 304 (if possible -- size of the furnace often becomes a constraint) at a bright-red heat to dissolve the carbides and then water quench to prevent their re-nucleation. Any of these techniques can be effective, but the additional cost has to taken into account.

Dibyendu De

Thursday, September 11, 2008

What Engineers Should Learn from the Big Dig Tragedy - 2008-06-16 10:47:00 - Design News

What Engineers Should Learn from the Big Dig Tragedy - 2008-06-16 10:47:00 - Design News


A failure analysis case study on failure of polymers and epoxy and their failure modes.

There are many lessons to learn.

Just click on the link above to read the whole story.

Dibyendu De

The Case of the Acid Test - 2003-02-17 00:00:00 - Design News

The Case of the Acid Test - 2003-02-17 00:00:00 - Design News

An excellent case study on Failure Analysis of Stainless Steel drum containing nitric acid of 90% strength.

a case study on Forensic Metallurgy

Dibyendu De