When I was in Eltam factory, and said the factory engineer about the xenon filled T8 Eco tubes, that are very sensitive to operating temperature variations and have reduced performance when the environemnt temperature isn't suitable, he mentioned the long T5s as having the same problem with reduced performance at lower temperatures, but above all, he said that the long T5s (HO and HE) have a rated operating temperature of 35degC (95degF).
Being have a rated instead of optimal operating temperature, meaning that lighting manufacturers had intentionally designed them, to have their best performance, much above the normal room temperature (rated tempearture of 35degC and 95degF, vs 20-25degC and 68-77degF of the room themperature range), in order to reduce their performace and short their lifespan, at normal operating conditions (Lamp life and lumen output, both drops at low temperatures).
So it is actually possible to make a long T5 (HO or HE) that will have the same peak performance temperature of the T8s and the T12s (Room temperature).
This made me the suspecting, that the "35degC rated temperature" of the long T5s, is actually a built in planned obsolescence inside the long T5s, wich accounts for their very short life.
What do you think about this?

The problem is the other way around: The T12 and T8 were rated for 25degC, but mostly operated at 35degC and above. So all the T12 and T8 rating was too optimistic compare to the real life. When the lamp is rated for 35degC, it way more correspond to the reality of the most installations.

Majority of fixtures heat up compare to the ambient air. But it is the warmed up environment, what become the "operating ambient temperature" for the lamps. And after few years of fluorescent use it became obvious, than keeping this temperature low, so lamps would still retain their performance, would be the major fixture design challenge, what was never successful except for really open fixtures.
So most of the T8 are operated way above their rated temperature.
The lamps were still rated for 25degC free air, as it was the only widely available accurate reference. At early days all lamps used only the liquid mercury, so have the same thermal characteristics, so for benchmarking purpose it didn't matter, if the real operating temperature was different. So the 25degC standard was set amd it pass even into the IEC.
When the technology evolved, it became possible to tune the lamp design, so the optimum temperature could be changed. But the marketing required the lamps look as good as possible on the figures in the datasheet, so the lamps were tuned to reach best figures on paper (so according to the standard), rather than to reach the best performance in service. Nobody in the lamp industry cared about the impossibility to reach that.

What came with T5HE/HO was the ability to fine tune the optimum temperature to virtually any desired value. And with that came the focus to optimize the lamps for real mission conditions, so allow the fixture designers to actually meet that optimum temperature with real products. So the 35degC was found as the best compromise: Not too high, but still manageable with both fixture and lamp designs.
But then the European Union came with strict requirement to still rate the lamps according to the old standard. They don't cared it was ridiculous, it was standard, so makers had to use that for rating. And as many makers would came with "better T5's than the major brands" by optimizing them for 25degC (so they would look better on papers) and create an unfair competition (as their product would be inferior in the real life), the standard was finally corrected to use the realistic 35degC instead of the ridiculous 25degC.

Of course, there is a draw back: In open fixtures the temperature is lower. But that is something the fixture designer could easily correct (add a glass sleeve,...). But the opposite (lowering the temperature in an enclosed fixture) is simply not possible at all.

Ok I understood. T5s really can get optimal performance in closed fixtures.

Of course, there is a draw back: In open fixtures the temperature is lower. But that is something the fixture designer could easily correct (add a glass sleeve,...). But the opposite (lowering the temperature in an enclosed fixture) is simply not possible at all.

Is this possible to overcome the problem with the higher optimal performance for long T5s, in open fixtures (Battens for example), by covering the lamp cold chamber (The end with the longer electrode, where the mercury condenses there), with a sealing tape (Generic name: Masking Tape), or an adhesive tape (Generic name: Sellotape), instead of using a plastic or glass sleeve around the lamp?

Theoretically yes, but each lamp type from each manufacturer would have to be tuned again. The reason is, the surrounding air and not the cold spot temperature is defined in the standard, so one maker may use hotter running cold spot with more heating (e.g. shorter stern), while another colder running cold spot with less heat provided (longer stern), so each would require different insulation.

The way to go is to put transparent sleeve around the whole tube with air between the sleeve and tube and design it's diameter so. the air below the cold spot is at the rated temperature. The air gap between the sleeve and tube would ensure the air temperature won't differ with lamps of different design.
I've seen T8 or even T12 sleeves on many installation, so I would guess the T8 sleeves would be around the optimum for HE and T12 for HO lamps...

This is a good idea. If they were frosted tubes you'd also reduce the glare from these thin tubes!


BG

T5's last forever, longer in fact than most T8, given that they're both operated on preheat or programmed start ballasts. I have a 46" 28w Normal output T5 that has well over 14,000 hours on it and still works. They also have excellent lumen and spectral maintenance over T8 and T12. T5 still hasnt become popular here in the USA in the commercial lighting sector due to the inability to retrofit existing T12 or T8 fixtures due to lamp length incompatability, but I have definately seen the benefits of T5 over T8 from my own experience. T5's here are available in NO and HO, what is HE?

@BG:
On the installation I've seen they were indeed frosted. Clear sleeves around "frosted" nature of thin tube would look odd in an open style fixture.
The covers played another role: They hide the dark ends of the T5's, they got gradually dimmer towards the end, what does not look as strange.

@BlueHalide: All well made fluorescents last forever for home use (10 khours really mean 10 years with the average 2.7hours/day pattern)
The problem with T8 is, the cheepeese makers pushed the price so low, than even major brands had to cut corners in order to stay on the market. Because many customers decide only based on price tag alone, that is fact the makers can not ignore...

The T5 are still made quite well, because they have still quite good price on the market, so there is sufficient budget to make them still well.

Osram Germany FH 28W/8xx HE T5, which operated in a Gaash Sealight Plus sealed fixtures, have an average live of several months of continuous operation, on a Helver multiwattage T5 HE HF ballast (Preheats their electrodes for 1/2 secs, despite programmed and not PTC start), in the corridor of the storage of Carmel hospital (Except a 3 secs power interruption at sunday at 7:00). This is my experience with them at least.

@dor: Then there is something wrong with the ballasts. I haven't seen any tube type failing so early, unless it was abused (inappropriate ballast including some of my experiments, ridiculously frequent switching on ballasts not designed for that,...).
If the operation is continuous (10 hours and longer per start), the starting method does not matter anymore, as the hours count up way faster than the number of starts, so even on a simple, instant start ballast the lamp should make in average way more than the rated life...

And if your observation is based on one failed tube, it was most likely bad luck for a defective piece. Don't forget, the rating is for median survival, so after the rated life (include the reference ON-PFF pattern), exactly one half of the large quantity of tubes are supposed to be dead.
I just checked the Osram website, after 2000 hours (what is about three months of 24/7) 5% of the tubes are expected to fail. So if your installation have 100 tubes, 5 failures are exactly spot-on the official manufacturer's rating.

And if the ballast was dead as well, it was most likely the ballast, what failed at first and as a consequence killed the lamp.

Medved: See this video , of a Helver programmed start HF ballast for 14W, 21W, 28W and 35W T5 HE, how it starts an Osram Germany FH 28W/865 HE T5 (In this case, the lamp is EOL), at the corridor of the storage of Carmel hospital, after a 3 secs power interruption, because of discharge testing of the emergency generators.

Medved: See this video , of a Helver programmed start HF ballast for 14W, 21W, 28W and 35W T5 HE, how it starts an Osram Germany FH 28W/865 HE T5 (In this case, the lamp is EOL), at the corridor of the storage of Carmel hospital, after a 3 secs power interruption, because of discharge testing of the emergency generators.

On the video is classical programmed start, but as I wrote, when the lamp is continuously burning for long time per each start, the starting method does not play any role in the overall lamp life.
The starting treatment become important in applications, where the lamp is ON for short time only, so it would run into very high number of cycles pretty early in it's life. But this isn't your case (according to your description)
Only the lamp does not look like failing yet, if the colors are correctly displayed and the ends are really white (from the discharge) and not orange (from the Planckian radiation only)

Medved: I haven't seen any white colour of a thermionic arc on the ends of the lamp, when it was preheated. Only orange colour from the Planckian radiation of the electrodes. Also, the lamp had a heavy blackning at the ends, when it was off.
A rectification of a fluorescent lamp on an electronic ballast, is invisible to the naked eyes (Because of the HF operation), and can only be detected if the lamp sufferes mercury migration, and for this to be happen, the lamp should be rectify from several minutes to several hours. Mercury migration, don't occurs instantly when the lamp rectify as soon as it turned on.

With most electronic ballasts of today (so with series DC blocking capacitor), there would be no mercury migration at all, as the current would have no DC component.
The mercury migration is caused by DC current (it is in fact plasma electrolysis), so when the current have no DC component, the gaseous mercury won't have any reason to migrate.
What the rectification cause is only the DC voltage component and that is detectable only electrically (and many ballast protections use that to detect lamp EOL)
The only signature of rectifying lamp on most ballasts may be only the orange glow from one of the filaments. And it is not a signature of rectifying lamp directly, but of failing cathode, what then usually lead to lamp rectification. But the rectification itself you may only guess (without electrical measurements).

I recently had to replace a ballast that drives a single F54T5HO, the replacement ballasts are expensive, much more than T8 ballasts. So I purchased a 2-lamp T8 ballast (2x F32T8's) and wired all the outputs to the single T5 lamp, so that lamp should be operating close to its rated wattage. It seems to work, and the lamp appears at normal intensity as it was with the old T5 ballast. Is there any problems with doing this?

I recently had to replace a ballast that drives a single F54T5HO, the replacement ballasts are expensive, much more than T8 ballasts. So I purchased a 2-lamp T8 ballast (2x F32T8's) and wired all the outputs to the single T5 lamp, so that lamp should be operating close to its rated wattage. It seems to work, and the lamp appears at normal intensity as it was with the old T5 ballast. Is there any problems with doing this?

If you were counting as "32+32=64, what is close to 54W", it is and quite severe problem with that methodology, because this math simply does not work with any discharges.
The voltage is dictated by the lamps (it stay pretty constant over wide range of the supplied current), so if the ballast is able to deliver it, the lamp light, if not, the lamp does not light at all.
Now in order to feed it on rated wattage, you have to supply it by a current source (in ideal form a device feeding constant current into whatever load voltage) at rated current. Using ballast of higher current mean you operate the lamp at higher wattage, so it may overheat.

Now the ballasts are designed to be very close to such current sources (at least around the desired operation range), so if a ballast is designed to feed 0.46A into 110V load, it would feed about the same current even into 160V load (so higher impedance). But the transferred power in the second case would be nearly double and that may overload some of the internal circuits, so it may fail earlier.
And here the ballast see the arc voltage as it's load, so even when the connected lamp is of lower wattage than the ballast label, the higher arc voltage cause the ballast's constant current output to deliver higher real power than the ballast rating. So with that, you could easily overload both the lamp, as well as the ballast.
Other problem could be with lamp ignition: The ballast is able to generate only limited voltage for lamp ignition and it should suffice for the supplied lamp, otherwise it won't work. Some ballast rely on the lamp ignition voltage to keep clamp the voltage to safe (for the ballast) level. So if you connect a lamp with higher ignition voltage, the ballast would encounter overvoltages during startup and it may fail soon.

Now if you want to operate different than rated lamp, you have to first check, if the lamp rated current is about the same as the rating of the lamp the ballast is rated for.
And then the second, you have to check, what cause the arc voltage difference to the ballast and overall system, behavior:
- Higher lamp voltage than the ballast is designed for mean the ballast is overloaded. If it uses the EOL protection based on the arc voltage, it could trigger and shut the ballast down. But the protection have some margin designed in, so it allow the ballast to be overloaded, while still not tripping. It is designed to protect against lamp EOL (so the arc voltage rise quite quickly - in few hours, so the ballast is not exposed to the higher power for so long), not against lamp vs ballast mismatch (the overpower is there for long time)
- The lower lamp voltage, mainly the ignition voltage, could cause premature lamp ignition during preheat phase (on programmed start ballasts) and that would make the preheating ineffective to protect the electrodes from the starting wear and some controllers may get fooled so, they start to behave erratically.

Now when connecting outputs together: That could work ONLY, when both the lamp circuits come from common inverter stage, as you may connect together only outputs, that are guaranteed to be exactly on the same frequency and in phase. And it should be parallel branches as well...

With your twin F32T8 ballast and single F54T5HO lamp you seems to be lucky, because the F32T8 multilamp ballasts do usually have one common inverter (so the frequency and phase condition s are met) and the F32T8 rated current (0.23A) is about half of the F54T5HO (0.46A) and the F54HO have a bit lower arc voltage, so it won't overload the ballast, but still it won't be as much OFF. The only problem could be with the ignition voltage: The thin long F54T5HO could require higher voltage for ignition (about 1200V peak with electrodes preheated) than the F32T8 (600V even with cold electrodes), so I see the danger for the F32T8 ballast, where the lamp could allow higher voltage than the ballast is designed for and so the ballast may fail (the resonant capacitor could have insufficient voltage rating for the voltage the lamp allow)