Anyone familiar with HAST (Highly Accelerated Stress Test)?

[reynald]

Hi guys,
Im not into reliability so this may be a very very basic question. If you are familiar with the Temperature&Humidity Oven/Chamber, how do you use it to gather failure times? IF for example I have a cycle set to 600 hours, do i have to break the cycle at regular intervals to check if a sample failed and then continue the cycle until 600 is met?

Regards,
Reynald

 

[Stijloor]

Re: Anyone familiar with HAST?

Hi guys,
Im not into reliability so this may be a very very basic question. If you are familiar with the Temperature&Humidity Oven/Chamber, how do you use it to gather failure times? IF for example I have a cycle set to 600 hours, do i have to break the cycle at regular intervals to check if a sample failed and then continue the cycle until 600 is met?

Regards,
Reynald

Highly Accelerated Stress Test (HAST)?

Stijloor.

 

[Miner]

Re: Anyone familiar with HAST?

Hi reynald,

What are you trying to do with the test?

As Stijloor indicated, HAST is Highly Accelerated Stress Test. This is another term for HALT (Highly Accelerated Life Test).

The purpose of this test is simply to break the product, determine the mechanism of failure (Physics of Failure), improve the design and repeat the cycle. The purpose of HAST/HALT is NOT the time to failure, it is simply the failure itself.

If you are really interested in the time to failure, you have four options:

• Demonstration test: Test parts to a specific time with a predetermined maximum number of failures to demonstrate a specified level of reliability
• Estimation test: Test parts to failure to estimate the reliability level
• Accelerated Life Test (ALT) test: Test parts to failure at multiple levels of stress to estimate the reliability under normal operating conditions
• Degradation test: Same as ALT testing, except the degradation in the performance of the product is used to estimate the time to failure rather than testing to failure.

For demonstration testing, you can run the full 600 hours. For the other tests, you will either have to break the cycle at intervals to assess failure (called interval censored data), or monitor the product real time to define the exact time of failure.

What type of product are you testing?

 

[reynald]

Thanks guys,
I am asked to create a life model and if possible find ways on how to accelerate the testing since we can not afford to wait looong hours to have the results. I was the one asked because I was the only one with background (i mean i had a brush with it during college, not really practical experience) with life data analysis. My problem now is how to gather life data to be able to estimate the distribution of failure times. By failure i mean a corrosion took place in a piece of an alloy. We do stress test by subjecting the product to varying humidity and temperature by loading it to a HAST oven. I can not monitor the product in real time, but need to have it inspected under a microscope to see if corrosion took place. In this case is this the direction i need to take?

..you will either have to break the cycle at intervals to assess failure (called interval censored data)

 

[Miner]

Do you already know the accelerating relationships of temperature and humidity? If not, read up on it at Weibull-dot-com.

You may need to perform a reliability DOE first to establish an acceleration model specific to your situation. This will define the main and interactive accelerating effects of temperature and humidity versus time. Test each run to the 600 hours, stopping every 100 hours to measure the amount of corrosion of each experimental run. Design the DOE for 3 factors (temperature, humidity and time). Note: this is a longitudinal repeated measures design, so the analysis is slightly different from a standard DOE.

Once you have this relationship, you can use it to predict the corrosion under any conditions based on the accelerated corrosion. You can also use the amount of corrosion at earlier times to predict the corrosion at later times or the time to an unacceptable level of corrosion.

 

[reynald]

Thanks Miner,
This is a great advice and the site is a very good read. Now I have a direction to take. I will consider running a DoE to understand more what is happening and use the resulting model as a starting point for improvement items.

Thanks again,
Reynald

 

[DarrellH]

Hi,
I think you need to consider "salt spray / fog testing" this involves (not surprisingly) taking a sample and spraying it with concentrated salt solutions, in a test chamber, dependant upon solution strength and test duration this is extrapolated to give "normal" atmospheric life.
this is commonly used in the automotive supply chain, ASTM B117 is the standard.

You can examine the sample during the test cycle and re-commence testing as required.

 

[Mikishots]

Hi,
I think you need to consider "salt spray / fog testing" this involves (not surprisingly) taking a sample and spraying it with concentrated salt solutions, in a test chamber, dependant upon solution strength and test duration this is extrapolated to give "normal" atmospheric life.
this is commonly used in the automotive supply chain, ASTM B117 is the standard.

You can examine the sample during the test cycle and re-commence testing as required.

The results should not be dependent on solution strength for ASTM B117. If it is, the test has been performed incorrectly.

 

[DarrellH]

The results should not be dependent on solution strength for ASTM B117. If it is, the test has been performed incorrectly.

Ok, you are correct in that, but, (being equally pedantic) I actually said solution strength is a factor used to extrapolate results.
Anyway, the original questioner was asking about using HALT to test corrosion on an alloy, I was merely trying to point him to ASTM B117 as a more relevant option - Agree?

 

[Mikishots]

Ok, you are correct in that, but, (being equally pedantic) I actually said solution strength is a factor used to extrapolate results.
Anyway, the original questioner was asking about using HALT to test corrosion on an alloy, I was merely trying to point him to ASTM B117 as a more relevant option - Agree?

Nope. The OP (if that's what we're going to use) was also talking about a stress test involving temperature combined with humidity. B117 specifies a specific chamber temperature, and isn't an option - it can only do part of what he's currently doing now.

I re-read your post - you've stated "taking a sample and spraying it with concentrated salt solutions, in a test chamber, dependant upon solution strength and test duration this is extrapolated to give "normal" atmospheric life." Solution strength is not to vary, so results do not depend on it. I would also be curious about extrapolation to "normal" atmospheric life; it's not possible to correlate hours in the chamber into a unit of time in a natural environment.