Hi Folks,

I'm in the midst of modifying my procedure for measurement of pump flow and am seeking some advice on acceptability specifications. First some particulars: The pump is a Waters 600E with 100uL heads that is about 10 years old, has been in constant use, but has been well cared for and recently rebuilt (last week as a routine PM by Waters techs with Waters parts). It holds pressure very nicely. (Currently with a 150mmx4.6mm C18 column at 1.0 mL/min, I'm seeing between 900-924psi - not bad). Our current procedure for measurement of flow rate involves doing so without the column attached to the system. Under these conditions, the pump gives an indicated flow rate very close to the indicated rate - no more that 2% error at 0.5, 1.0, and 2.0 mL/min. With the column attached, however, the flow rates are somewhat lower than indicated. 1.0 mL/min indicated is actually 0.93 mL/min and 2.00 mL/min indicated turns out to be 1.88 mL/min. In modifying the procedure to be more like a real-world world situation, do you think it would be OK to loosen the acceptability criteria a bit to accomodate the results of the modified procedure? I see no leaks anywhere in the system and am having no problems with retention times being any greater than they ever were, so I'm inclined to make the changes to the procedure and the acceptability criteria and not worry about it, but I'm interested in your thoughts on the subject and whether you think these numbers are reasonable or not.

Thanks again!

Chris

I think you should fix the problem and not the specifications. You are getting about 6.5-7% variation from the desired flow. Specs are usually around +/-1%. You could try using water as a mobile phase. It does not compress at high pressures like ACN or MeOH. You should run your procedure at pressure, at least 300psi, to make sure the check valves are operating correctly. It seems that they are not operating correctly since they are not holding pressure and therefore are not delivering the correct flow. I believe Waters does, or is supposed to do, testing to make sure the instrument works after the PM's are performed. If you find that it does not work after the PM, call them back and have them fix the problem under warranty for their work.

OK, a few clarifications...

First, I did use water as a mobile phase for precisely the reason you describe.

Second, Waters' procedure as it appears on the PM sheet supplied by the Waters tech reads as follows (exactly):

Calibration protocol:

Flow Rate

1. Connect a short length of 0.009 in tubing to the outlet of the pump.

2. Prime and flush the system with Methanol.

3. Using a 10cc graduated cylinder, measure the flow output of the pump at 1 ml/min for three minutes. Start the pump flow, allow the cylinder to fill to the 1 mL mark, and start the timer. Stop the timer when the solvent reaches the 4 mL mark on the cylinder. The time to fill the cylinder chould be correct within 9 seconds (between 2:51 and 3:09 min)

Now, this looks to me like a +/- 5% window with a pump that isn't under any load. My pump will pass this test easily. Using a longer (10 min) variation of this test with water as a mobile phase rather than methanol yielded a flow rate of 0.998 mL/min as measured by weight (on an analytical balance with water density corrected for ambient temperature).

The reason I wondered about changing the specs is that I've changed the conditions under which the flow rate was measured and one might conceivably expect slightly different results, which is why I asked the question. If I had NOT tried a revised procedure and got low flow rates, I wouldn't even be here asking the question - I'd be troubleshooting the system, but the test I ran was different from both the one Waters describes and the one this lab has used in the past.

I'm not looking for a cheesy way out, I'd like to know whether any of you have observed this on a very recently rebuilt, well cared for pump that appears to hold a prime well and gives similar chromatographic results as it did prior to rebuild (it appeared to be running OK then, as well).

I'm going to call my service tech, but am interested in the comments of the folks here.

Thanks,

Chris

I'm also currently revising HPLC OQ/PV procedures. We run our flow precision and accuracy at pressures of ~750 and 1500psi. Our accuracy spec is 5% and precision is 0.5%. We take ten replicate readings using a calibrated digital liquid flow meter (Humonics 100 from Altech). We have 0-5 y/o Agilent 1100s and Waters Alliances.

I have not observed better accuracy at lower pressure, I also suspect a check valve problem. Good luck.

Thanks for all of the input! I believe that I've isolated the problem, though, and it is not the check valves. When I did the test with the column attached the first time, I did so with the column in it's normal position at the outlet of the autosampler. I did this in an attempt to best simulate real-world conditions. This yielded poor flow readings. This morning, I attached the column directly to the pump outlet and measured the flow a second time. Results were fine: 0.50 mL/min indicated was 0.495 mL/min, 1.00 mL/min indicated was 0.988 mL/min, and 2.00 mL/min indicated was 1.96 mL/min. So I went back and attached the column to the output of the autosampler, remeasured at 1.0 mL/min, and got a reading at 0.920 mL/min. So, I most likely have a leak in the autosampler, which will be the next avenue of pursuit. Interestingly, had I simply evaluated each piece of instrumentation in isolation, I might not have found this for a little while. If the leak is on the input side of the autosampler, it might conceivably have passed injector accuracy and precision tests, the pump is obviously OK, and my detector is dandy as well. I'm curious as to how you guys came to the check valve theory, though. I have never seen a pump with marginal check valves deliver constant pressure. In every case of bad check valves I've seen (and I admit they are few), the pressure transducer tells the tale. Have any of you found otherwise? If so, under what circumstances?

The only caution I have is that Waters 600 have a tendency to have poor functioning pressure transducers over time. It may read one thing but be another

Thanks, I'm actually quite aware of that - the transducer in the pump in question is five days old and appears to be functioning well. I have found that while they're not the most accurate beasts out there, often they still will give a reasonable indication of pressure variation between the heads. Oh, and I found the autosampler leak too and it's a big one. Apparently the seals in the injector (newly installed by Waters) are faulty or the installation was incorrect. I had hoped for a lousy compression fitting - no such luck. Service engineer got another call on that as it wasn't leaking before the PM. I always feel that if it ain't broke, don't fix it, but we've got an annual rebuild schedule to hold to, so whether they need it or not, all machines are redone. Usually this is not a problem, fortunately. Thanks to all.