Given the discharge pipe size I suspect that your sampler is far too small:  I have used a hand sampler as you describe on tailings but only upto an 8" pipe.  As sampling needs to be fairly precise I'd suggest you contact an expert in sampling and would recommend (even if you are not in North America) : Ricardo Herzog
Email rherzogheathandsherwood.com at Heath & Sherwood (1964) in Kirkland Lake Ontario Canada Toll Free (North America only):1-888-467-2675 (1-888-GO SAMPLE)

The turbulence in your pipe should prevent stratification, but with a 30" pipe and only being interested in pulp density you may find it easier, and perhaps cheaper than sending an operator to take a sample, to consider an automated density measuring device,   perhaps a nuclear density meter?.  If you continue with hand held sampling you may need a primary splitter on the discharge pipe then take manual cuts from that reduced flow.

What data do you want to obtain from it and to what purpose? Pulp density %solids only, to run your system? Anything else?

Thanks for the replies. We have nuclear gauges but we take samples manually every two hours to ensure they are not deviating. This has happened several times. We do checks with a Marcy gauge and oven dry and if needed we need to adjust the density gauges accordingly. 

We occasionally do SG and PSD on the sample collected, but not always.

By how much is it deviating now? Do you have a photo of your nuclear gauge installation?

https://www.911metallurgist.com/blog/sampling

The data you get from the sampler shown and be the method you describe will be virtually meaningless.  I attach a file which covers practical aspects of sampling streams such as you describe.  It contains pictures of the correct sort of design.  The sampler should have an opening which is uniform from front to back.    In your case it will need to be long enough to cut the whole stream from front to back as you sweep across at uniform speed.  You need to cut the whole stream width in one sweep.  If you need more sample then take more than one cut or go slower.  The width of the opening and the speed will determine the amount collected in a single cut.  I would not rely on a spot sample but would increment a sample with six or seven cuts over the course of an hour.  The sampler opening needs to be suitable to the largest particle size (formula in attachment) and the sampler widens out below the opening to accommodate the total volume of sample.  With a sample taken as I describe you should be able to get meaningful data on not just pulp density but PSD, head assay, size by size assay, SG, etc. - the lot!

Many thanks for the replies. 

One issue is that the pipe is 30 inches with a flow rate of around 2500 m3/hr. Trying to cut a sample in one sweep is ideal, but not practical at this throughput. The images of the sampling in the PDF are very easy streams to cut. 

What data do you want to obtain from it and to what purpose? Pulp density %solids only, to run your system? Anything else? Can you clarify? 

Again, by how much does your nuclear gauge deviate now?

Honestly, if all you want is an operation %solids and the "sometimes" PSD/SG, the cutter you have if fine. Just be sure to sample constantly. Same angle, location, sample cutter, technique.  

You can spend $500,000 or $1,000,000 of a full pipe sample cutter and then what?  What do you gain by it?

A nuclear gauge will always vary. A variation in those has many causes. That's why you'll send out an operator to verify. 

Thanks David. I did mention this previously but we just do %solids to check the nuclear gauges to ensure they are not deviating. This has happened several times. We do checks with a Marcy gauge and oven dry and if needed we need to adjust the density gauges accordingly. 

The nuclear gauges can deviate around 2 - 3% with time.

OK. Just an operating density? 2 - 3% is nothing. Let it go. Tell me this:  Why do you think the nuclear gauge would read perfectly all the time? No matter the operating variations: PSD, SG, Tonnage.

Track your data, Adjust the nuke less frequently. Average it out and I bet the "error" disappears.

In a 1000 TPD plant, I am sure you will find other things to spend your energy on.

The problem is that we are restricted to meet a criteria and 2 - 3% is a large variation for us. Thanks anyway for the comments.

restricted to meet a criteria: what/why? You need to describe the problem are trying to solve. We may be able to help you without the $1M sampler.