Two weekends ago, after running up a long hill, my 1998 civic broke the radiator upper tank. First it was a very tiny puncture, so I decided to adventure and drive back home immediately (about 5 miles) after letting it rest 15 minutes. Less than 3 minutes driving downhill, it burst open the entire tank (suspicious, it already had a leak where it was releasing pressure), so I tow it back home.

Unable to get a new radiator as a whole, I took it out and dropped it in a radiator shop. They cleaned it and replaced the upper tank. The guy said the radiator was perfect. I installed the radiator back on the car, with a brand new Denso radiator cap, and proceeded on the regular bleeding instructions (as the Honda repair manual says).

After several unsuccessfully attempts to bleed it that way (it kept missing coolant), proceeded to bleed it with a funnel, but I was unable to get rid of the bubbles (even when it's just one every couple of minutes, they keep coming out). After warming the engine and letting it rest overnight, opening the cap always gives me an opening soda bottle sound (so its still pressurized). After several bleeding attempts, I suspected blown head gasket.

Went to a shop, and performed a radiator leak test (no pressure drop from 15 PSI in 15 minutes) and a compression test: got 150 PSI on cylinders 1, 2 and 3, and 151 PSI on #4. The engine runs smooth, no white smoke on the tailpipe, no water in the oil, no oil in the water, usual power and no overheat in normal driving conditions.

In total, 3 mechanics saw the car. One did the mentioned tests with me, other saw the engine idling and send me back home telling me I was crazy, and the other did the same after checking the thermostat (amazingly, none saw the bubbles on the radiator as an unusual thing). They all gave me false confidence that the engine was OK. Then I drove it in a massive traffic jam, just to see the temp gauge climbing half way to Hot. Turned off the engine immediately and towed it back home (again). The reservoir got filled to the top, but just this time.

I wasted a week looking for a shop with a block tester or cylinder leakdown tester, but I didn't find anyone, and actually several of them told me that “the leakdown test you're suggesting could harm the engine, your engine is fine, so don't do it”.

As I know this test is regularly performed on airplanes, I gave up looking for someone that knew about it and had the tool (I called everyone on the yellow pages) and build a test rig myself. I didn't build the entire two manometers and restriction in the middle device, but simply injected 70 PSI from a tank directly into each of the cylinders at TDC.

I saw the radiator bubbling on every single cylinder I tested (cylinder 4 more than the others). Is that usual? All the cases I've read about, only a single or a couple of cylinders leak. That fact makes me wonder.

To my knowledge the engine hasn't overheated a single time. I normally drive all cars with an eye on the gauge because I'm paranoid after I overheated a Cherokee a few years ago. In this Civic, the farther the gauge has climbed is half way between normal and hot, 3 times only: 4 years ago, original radiator gave up, two weeks ago after radiator burst, and driving on the traffic jam (and bubbling was already observed by then). On the two radiator incidents, the temperature was fine until they burst (and I found a safe place to park the car, then towed it back home).

I performed the test with a warm engine (I let it run a couple of minutes in the morning). I got TDC with a stick in the spark plug hole. The oil cap was on (as I was looking for air in the radiator).

Must admit the car had been running with filtered water for half a year before the radiator issue, as it started loosing coolant and the lowest I get all year round is 68 F (20 C). I started adding water instead of coolant. I promised myself to check for the leak, but was really small (less than ¼ of the reservoir each two weeks, and it just didn't reveal anywhere). Eventually, the coolant thinned out to almost clear water.

The radiator shop said the radiator was in great condition. The pump has less than 25000 miles and seems to be running good. The thermostat was checked twice (I performed one of those tests).

The odometer reads 182000 Km (113,000 miles). Automatic transmission. Engine never touched. The engine reads D16Y9. Seems like it was sold only in 3 countries, Venezuela being one of them. I believe the closest equivalent is a D16Y7, but without catalytic converter, O2 sensor and running off an OBD1 ECM.

I keep trying to burp it, but bubbles keep coming. The radiator gets pressured even after 1 minute of running the engine (from a dead cold engine). If I warm it to normal temperature, and then let it cool with the cap on, I get the soda bottle sound, and missing water. If I do the same process but without the cap, the radiator stays full. The reservoir level never changes (except for the traffic jam). I can't hear nor see air bubbling in the reservoir, not even after a few miles driving.

Should I proceed to tear down the engine, hoping that resurfacing the head and replacing the gasket would do it? (fingers crossed is not the block) Any reason why it could have failed? I'm afraid of not solving the original cause, just to see it fail again in days, weeks or months. I'm afraid to do all that work just to discover it wasn't the head gasket.