THE DEVELOPMENT bF A TRANSFORMEK COATING

                        TEST PROGRAM fOR SUBMERSIBLE URD UNITS

                                      Robert A. Pratt and Thomas J. Cronen

                                       Northeast Utilities, Hartford, Conn.

Summary



Many tests are available to check the physical properties of coatings but there are none to simulate the long

term environmental conditions encountered by the coated steel URD transformer. Based on research done

by others in the coating field, a test has been developed which will screen new and existing coatings and give

reasonable assurance of a long life expectancy.



Coated sample plates are placed in a universal solution, are aged by overstressing with a closely controlled

cathodic protection voltage and are graded on the basis of the comparative resistance of the coating to hydro-

gen disbondment and electroendosmosis. Added to this is Thermal cycling which provides a check on coating

resilience, sag and solubility. The test is performed in six weeks (30 days under cathodic protection) and many

samples may be placed at random in a common solution. From a practical design point of view, the physical

properties of the coatings should undergo standard tests; however, the combined effect of all these properties

is simultaneously evaluated in this single environmental type test.



Whether a selected coating can be applied on the tank with the same results would be determined by a full

scald tank test. The only remaining variable, then is whether the coatings are cured properly on an assembly

line basis. An effective quality control measure such as color grading of light colored coatings is a possibility.



The coating industry’s know-how and knowledge of transformer operational requirements is providing the basis

for developing sound coating Systems. The results are small scale screening and full scale

testing bear this out. We believe that superior coatings can be successfully developed, properly applied and

effectively cured on a production line so as to provide a long life expectancy.



INTRODUCTION



The decision by many electric utilities to bury new residential construction has presented some unique corro-

sion problems to the industry. Our own field and labortory investigations indicated that the bimetallic coupling

of a bare copper neutral to any exposed steel on a tranformer creates a severe galvanic condition which is

aggravated by heat cycling the transformer in a moist environment. Two field checks of unprotected coated

transformers showed 100% corrosion penetration of a mild steel transformer tank after 16 months�? service and

25% corrosion penetration of�? a stainless steel tank after two years�? service. The corrosion of the stainless steel

was not visibly apparent. A laboratory analysis of the tank was required to reveal the pit corrosion. The immedi-

ate recommendation was to place all transformers under cathodic protection, train field personnel in the proper

installation of equipment to minimize corrosive conditions5 and investigate tank materials and coatings. The

conclusions of our investigations are being reported in this paper.



Among the general performance considerations, the tank system must resist water absorption and degrada-

tion due to thermal cycling, provide heat transfer capability consistent with rated loading and have mechanical

strength to resist impact, impulse and short circuit conditions. Many methods are being used to negate the

corrosion problem such as: superior coatings on mild steel with cathodic protection, exotic metals, fabricaed

fiberglass tanks that eliminate bimetal couplings and finally, encapsulated core and coils which eliminate the

need for a tank. Since the ultimate solution will bear heavily on economics, we believe the superior coatings

development has the better chance for immediate success. Since there are kriown environmental aging, tests

applicable to pipeline coating systems, we followed this course.



The key to superior coating corrosion corrosion in the development of coating systems that have exceptional,

long lasting adhesive qualities. The biggest problem is to adequately represent field conditions for making ac-

celerated laboratory tests. We concentrated on developing a modified version of the Kish Salt Crock Test which

was designed basically for buried pipe coatings. The following describes our version of the test emphasizing

the theoretical as well as the practical aspects involved.



BASIC TEST DESCRIPTION



The test is an aging process created by applying a constant overstressing cathodic protection voltage across

the coating system through an electrolyte designed to simulate environmental conditions while alternately flex-

ing the eoatine by cycling the system temperatures (Fig. 1). The physical results provide a means for determin-

ing the relative rating of the coating systems.



ELECTROLYTE



The electrolyte is the “Gally Solution�? consisting of 1% by weight each of technical grade sodium chloride, so-

dium sulfate and sodium carbonate in distilled water.6,4 These salts were used by Gally because of the frequent

occurrence of these minerals in California soils. Subsequently, these reagents have been commonly found in

most soils in this country and provide a nearly universal test basis. Distilled make-up water is controlled with an

automatic float arrangement. The resistivity of the electrolyte is approximately 30 ohm cm. When the solution

resistivity is maintained below 100 ohm cm, the relative location of the anode, reference electrodes and test

specimens have considerable latitude. This allows the testing of many cells simultaneously.



TEST PANELS



Our test panels have ranged in size from 2 x 4 inches to 6 x 8 inches in standard gauges. A 1/4”diameter

holiday in the coating is centered an the lower half of one side of each coated panel and Is made with an en-

graving tool. This type tool is used to minimize any lifting beneath the holiday edge coating which may occur

with the twisting station of a large drill bit. An identical pair of sample plates is tested simmuItaneously in each

parallel electrical circuit. This eliminates the possibility of rejecting a good coating if one sample should not

perform well due to poor preparation or coating application.



CATHODIC PROTECTION



To simulate a protective potential, a filtered DC source is used to depress the samples to -1.5 volts DC rela-

tive to a CuCuSo4 reference. The magnesium anode is not used as a voltage source since solution potentials

on a good coating with magnesium anodes may vary from -1.55 to -1.40 VDC referred to CuCuSo4 and will

decrease if the coating starts to fail. We stipulate a constant voltage of -1.5 on all sample plates regardless of

local condition changes.



THERMAL CYCLING



When all preparations are completed for the test. the solution is brought to and maintained at 60°C. A constant

mild agitation of the solution assures thorough mixing. Daily current readings are taken and the potential ad-

justed if necessary.



The possibility for extended exposure to cold temperatures, as in outdoor transformer storage, is why we use

a low test temperature limit. Samples are placed in the freezer over the weekend only after the electrolyte and

samples are allowed to come to room temperature. This prevents extreme thermal shocks on the samples. The

plates are removed from solution, placed in the freezer at -15°C (5°F) ,placed back in the electrolyte at room

temperature on the next work day and heated to 60°C (140°F). Since the aging is based on 30 days under

cathodic protection, the period in the freezer is not counted as test tine.



A CuCuSo4 voltage reference cannot remain permanently in solution, therefore, a platinum reference elec-

trode is used. A calibration an the platinum was made to maintain a -1.5 VDC relative to the copper reference.

Comparison between the copper and platinum references was checked at different temperature levels (Fig. 2).

Since the error between 0°C, to 60°C was within 2.%, and less than 1% at the test temperature levels, correc-

tions due to thermal cycling is not felt necessary.

AGING THEORY1



A cathodic protection voltage in excess of the metal’s equilibrium potential is impressed on a coated metallic

surface to electrically stabilize it and prevent corrosion; the resulting stress placed on the coating has a de-

structive tendency. The ability of the coating to withstand this destructive force is the measure of its quality.



If the stress received during a number of years is Concentrated over a short period of time, the physical effects

of this test pexiod should be comparable to the long term effects and an accelerated condition is developed.

Based on previous studies it has been determined that for every 0.1 volt impressed above the equilibrium volt-

age of the metal the destructive force on the coating doubles. Therefore, assuming an equilibrium condition of

-0.7 VDC, an impressed voltage of -1.5 VDC has an acceleration aging factor of approximately 300, providing

the samples successfully approach the full test period. This figure was verified by comparing artificially aged

laboratory samples coated with coal tar enamels against aged field samples having the same coating.7

Coal tar enamel is a type of coating having a successful service life exceeding 40 years.



TEST VOLTAGE



There is a practical limit for overstressing sample coatings. Impressed voltages greater than -2.0 VDC create

coating failures which are difficult to evaluate with consistent results.4



A voltage greater than -1.1 VDC is beneficial for testing since under these conditions hydrogen gas is gener-

ated at the cathode. This gas formation at a holiday has a bombarding effect on the coating edge. If the coating

cannot resist this attack, the hydrogen gas will propagate between the steel and coating resulting in undercut-

ting. This phenomenon is useful for determining weaknesses in coatings. Based on the foregoing, a functional

voltage of -1.5 VDC is used for test purposes.



ELECTROENDOSMOSIS EFFECT



If a slight current is allowed to flow through the pores of the coating, electroendosmosis takes place. Electro-

endosmosis is the penetration by liquids, gas or polar particles through an imperfection in the coating which is

under the stress of a cathodic potential (Fig. 3).3,4 This is similar to “capillary action�? in that either the gener-

ated hydrogen gas or the electrolyte is forced through the coating pores to the metal surface. Coating defects

with slightly pourous areas will pass the hydrogen gas whereas the more pourous coating areas will fill with the

liquid solution. In either case, blisters will form which eventually cause failures and any blistered area is con-

sidered to have failed. The liquid within a blister will have a relatively high pH. However, alkaline conditions are

not corrosive to the steel surface.3



On occasion we have found some coatings which are soluble and precipitate particles on the test plates. This

is usually indicated by a sharp drop in current readings with a visible deposit collected at the holiday. This is

corrected by removing the sample causing the problem, reversing the polarity on the remaining samples until

the deposit is rejected and renewing the solution.



We prefer using a platinum anode since the deterioration of other type anodes may cause discoloration of the

electrolyte and any loose particles may collect at the cathodes. This plating action again is caused by elec-

troendosmosis and can have a retarding effect on the test. The stability of the platinum anode eliminates this

problem.



TEMPERATURE EFFECTS



In order to check the thermal effects on the test, a chart was placed in the circuit to record the current require-

ments while maintaining a constant voltage and increasing the temperature (Fig. 4). By raising the solution

temperature from 16°C to 60°C and maintaining a constant -1.5 VDC to CuCuSo4 the protective current re-

quirement increased by 40%. Since the amount of hydrogen gas generated at the cathode is dependent an the

current density, the condition for hydrogen disbondment becomes greater. The effect of electroendosmosis is

also increased through any porous area. 3



Heating the electrolyte will determine whether there is a sagging or disolving effect which would be harmful to

coating performance. Some coatings may perform better at higher temperatures than at lower temperatures

since the shrink stresses developed after coating application are relieved. On the other hand, at cold tempera-

tures, coatings may be less resilient and may crack or disbond by exceeding the shear of the band tensile of

the coating. These conditions are caused by the different coefficients of expansion and contraction of the coat-

ings and the base metal.



In actual field service, the transformer loading is cylic. A cyclic wet and dry condition an the plates creates a

more severe and realistic lab test. Therefore, we have been performing the tests by cycling the temperature in

a liquid and air exposure (Fig. 5).



In field service, heat transfer on a coating system is from the inside out. In the laboratory, the coatings are

heated from the outside. A check on the method of applying heat was made with a 4�? steel cylinder capped at

one end.



A supplier coated the cylinder just as he had coated sample plates. The

cylinder was partially submerged in the salt crack and tested. Heating an oil solution placed within the cylinder

simulated a more realistic directional heat transfer through the coating. The test results were identical to those

obtained with the sample plates (Fig. 6). Subsequent full scale

tests have verified the consistency and reliability of the test panel screening tests (Figs. 7,8).



EVALUATION



Our basic concern in evaluating transformer coating systems is how well they perform and for how long. Per-

formance is judged by comparing tested samples. Presumably, the better performers will provide a Longer life

since the test conditions are accelerating realistic, reproducable and the results are consistent.



Coating samples should go a minimum of 30 days in the electrolyte under the specified cathodic potential. Al-

though a part of the test, the removal of the samples from solution and placement in the freezer does not count

as test time. A high current change or excessive blistering (25%+ of submerged area) will warrant an early test

termination.



Failure due to environmental aging is detected either by propagation of gas under the coating film at the

holiday or by liquid or gas penetration through the coating causing the formation of blisters. A needle probe1

or similar device may be used to check the effective coating contact at the holiday or any other areas which

appear disbonded. A poor edge coating will also become obvious during the test. The percentage disbond-

ment relative to the total submerged area becomes the major factor for ranking coating systems, providing the

general physical appearance has not changed drastically. Comparison can be made with the upper portion of

the test plate which has not been submerged in solution.



Additional physical limitations become obvious after the test. Coatings may shear at the edges and peel in

sheets due to the lack of resilience. Others may change their appearance and physical characteristics due to

the loss of plasticizers or additives which may cause brittleness. The cause and prevention of failure may be

corrected by a complete physical inspection of the tested plates.2



Daily current readings are a good indication of local cell performance. Since the voltage is constant, any

change in current theoretically is related to exposed metal area. Any sharp increases in current reading would

indicate an impending coating failure. Any sharp decrease in current reading would indicate a possible deposit

of foreign material at the holiday. It should be noted that there is a normal gradual drop in current readings

at the beginning of the test during polarization of the holiday. We have determined that a high percentage of

disbonded area is not necessarily associated with a high current change. We have found complete sheets of

coating sheared away and just clinging to the metal with little current change. Blistered areas will not neces-

sarily create high current changes although they are considered failed. For these reasons, the use of current

readings in evaluation is limited.



A FULL SCALE TRANSFORMER SALT CROCK TEST



The application of a coating to a transformer on an assembly line basis can

be very demanding. The tank surface must have a proper anchor profile. The grit blast material and tank sur-

face must be clean, dust and oil free. The tank contours and edges may be difficult to clean and coat while the

cure time is critical. All these factors enter into the effectiveness of a coating bond.



We have developed a full scale transformer test for checking the application

and performance of previously screened coatings (Fig. 9). The test is essentially a large scale salt crock test

on a coated tube or fin type transformer tank.



The transformer tank is placed in the center of a 3�? diameter x 4�? high plastic container. Tap water is used for

heat transfer in the transformer and

the brine solution is placed between the transformer and the liner. A floating heater is used to heat the tap wa-

ter and cooling coils are placed in the brine. Temperatures are cycled from 0°C to 60°C with the same frequen-

cy as the screening test. The test is continuous for 30 days. The only changes in test procedure are that one

inch square holidays are arbitrarily placed at any critical areas such as welds, edges, etc. A,magnesium ribbon

type anode is used. A constant voltage DC source maintains the 1.5 VDC to CuCuSo4, cathodic protection volt-

age on the tank (details may be found in appendix B).



At the end of the 30 day test, the transformer is theoretically aged 25 years. A critical evaluation of the tank can

be made for acceptance or possible improvements.



COATING CURES



Although transformer coatings ran be screened and tested for coating application, there is no assurance that

the coating has been properly cured on a production basis. There are individual transformer tests that can be

applied for checking the degree of cure; however, this is costly and impractical.



A more realistic approach to this problem may be in color grading of light colored coatings. This can be ac-

complished at a well lit assembly line check point. For example, a light green coating may appear apple green

and grainy if it is undercured and brownish green if it is overcured. This method of visual inspection for properly

cured coatings is presently being successfully applied to other products.

ACKNOWLEDGEMENT



The authors would like to express their appreciation to: George Kish of the Dresser Manufacturing Company

for his counsel and encouragement, Floyd Norton for the unrestricted use of The Connecticut Light and Power

Test Department Facilities, Bill Osborn CL&P for the lab assistance, Dan Stanton CL&P for the full scale ther-

maI cycling installation.



REFERENCES



1. A simplified Method of Evaluating Underground Coatings, G.D. Kish, AGA Distribution - Transmission Con-

ference, Minneapolis, March 5, 1965.



2. Causes and Prevention of Coating Failures, NACE Task Group T-60-22, Materials Protection, March, 1970.



3. Four Phenoemna Affecting Cathodic Protection and Corrosion Rates, F.W. Hewes, Material Protection,

September, 1969.



4. Testing and Evaluation of Coatings for Underground Service, G.D. Kish, AGA Distr:ibution - Production

Conference, Philadelphia, Pennsylvania, May 8-12, 1961.



5. A Visual Training Aid for a Submersible URD System, R.A. Pratt, E.E.I. Tramsmission and Distribution

Committee, Phoenix, Arizona, January 20-23, 1970.



6. PreliEimary Laboratery Evaluation of Pipe Coatings by the Salt Crock Test, S.K. Gally, Pacific Coast Gas

Association Proceedings, April, 1949.



7. G.D. Kish, Dresser Laboratories.

FIGURE COPY



FIGURE 1 - Apparatus for hot salt crock screening test (detail Is in Appendix A).



FIGURE 2 - Voltage reference variations caused by variable temperatures



FIGURE 3 - Electroendosmosis effect - Left, disbandment caused by penetration and propagation - Right,

dark discoloration around holiday due to accumulation of metallic particles on the coating.



FIGURE 4 - Recording of protective current increase due to temperature increase while holding a constant

voltage.



FIGURE 5 - Identical plate coatings after salt crock test - Left, temperature cycled - 15°C to 60°C - Center, con-

stant temperature 60°C Right, room temperature. The left sample is grainy and blistered while the two right

samples are in excellent condition.



FIGURE 6 - Identical coatings on plate and inverted container. For comparison, the plate coating was heated

through the electrolyte while the container coating was heated from within the container. The results are the

same. The dark vertical streak on the container is a stain.



FIGURE 7 - Identical results obtained from the screening test andthe full scale tank test. Both indicate propa-

gation at the holiday.



FIGURE 8 - Note the characteristic blisters from the screening test and the full scale tank test. Both are

caused by coating penetration around the holiday.



FIGURE 9 - Apparatus for full scale salt crock test