Pressure Drop In Parallel Filters

[BenjaminM]

Hello again, I hope you all are doing well!!

I have a question regarding pressure drops in parallel systems.

Our current arrangement is as such. The process line pumping from the reactor tee's branching off to two vertical pressure leaf filters. These filters are used to remove sodium chloride crystals from our solution of solvent and product. The outlets of both filters are piped together. From there the flow goes to the next vessel in our process.

The current filters in place are roughly the same size. They are by no means identical though. One filter has about twice as many filter leaves as the other.

We have attempted to measure pressure drop across the filters with differential pressure (DP) sensors, though because of what I would consider a poor arrangement and faulty sensors we have never gotten a good reading on these filters. Additionally,I have no means to measure individual flow rates though each filter.

The need to know the DP across these filters is two fold. First as our filter cake builds up on the filter leaves, the DP will increase. This would be a good indication all is well or if the DP gets to high, that the filter is plugged or that our product is falling out of solution. Secondly if the DP were to drop it would suggest that a rip or tear in the filter mesh has formed and sodium chloride is not being removed.

Just for the sake of argument, ignore the effects of all valves, and piping leading too and from each filter. There are several feet of pipe between the two filters, a few 90's and full port ball valves. But again, only for the sake of argument please only consider the effects of each filter.

It seems to me any two points on the inlet side of each filter are under identical pressure at all times. Similarly any point on the outlet side of the filters are under identical conditions aswell. Should a tear form in filter A, filter B's DP would drop as filter A's does with the increase flow. Similarly if filter A were to clog, flow would divert though filter B. In other words the flow should change to maintain identical pressure drops though each filter at all times.

If this is the case then I should not need to measure the DP across BOTH filters. Only on the inlet and outlet sides of one filter need be measured.

Is this reasoning correct? What little information I have come across seems to support it.

As I mentioned above, the current DP sensors are not to my liking. Our process connections are 3/4". The DP sensor has two, 5' long, liquid filled capillary sensing tubes that connect to diaphragms at each connection. The DP sensors are "made" by a local valve and instrument supplier. They leak at the welds, among other problems. I am thoroughly not pleased with them.

What I intend to do is have a 2" flanged connection tee'd into the line just before the inlet of filter A and just after the outlet of filter A. To each flange I intend to connect a pressure sensor with a 2" diaphram, and allow our PLC to subtract the difference of the two sensors to yield our DP. This does away with sense lines. Most likely I will go with Foxboro sensors, I have had few issues with them.

Is my logic correct? Does measuring DP across one filter offer a reasonable estimation of the DP across the second? Will comparing the outputs of two seperate pressure sensors stand to give a better indication that that of one DP with several feet of sense lines? Most ideally I would perfer one filter, properly sized for our application, however for the time being, I am stuck with two.

Thank you all for the time and help!!

-Benjamin

 

[davsy]

In my experience with pressure leaf filters (edible oil and biofuel bleaching and winterizing) where two filters are installed we have always used a duty/ standby system although there have been a couple of customers who have tried to push their flow by using both filters in parallel but we found the problem was we did not get even flow through the filters, we found this due to exactly what you are describing, the pressure drop across the whole system is shared between the two filters and as one starts to blind up the flow through that one slows as more and more flow diverts to the other filter which then increases in pressure due to the extra flow.

Our normal control set up is to use the inlet pressure or a set time as the point at which the filters are automatically changed over, unfortunately if you use both filters you only reach the "clean and change" set point when both filters are at that point, then it is a mad panic to clean down one of the filters to get the plant back up and running again. This is probably the primary reason for the duty standby set up.

We normally put a pressure transmitter in the line between the filter feed pump and the manifold where the valves route to either filter, we don't particularly worry too much about the outlet pressure as we are discharging through a cooler into an atmospheric tank and the cooler has a local gauge on the inlet to ensure there is no blockage in there.

 

[epoisses]

"Is this reasoning correct?" Yes, certainly. The delta P across the parallel systems will always be exactly the same, just like the delta Voltage across two parallel light bulbs will be exactly the same, even if the bulbs have very different resistances. No need to measure two parallel delta P's separately.

Davsy's approach (one on line, one blocked in, then change out at a certain delta P) sounds like the right approach to me.

 

[zdas04]

With filters of different resistance to flow in parallel, you will never get the same flow through both. Since dP is a function of flow rate and resistance to flow, you will get the same dP in the two filters, but you'll never know how much is flowing through each.

I've seen people apply the logic that "the filter with the higher flow rate will collect the most solids, so it will eventually be self-shifting to the other filter". I've never seen this work very well.

There are many valid reasons for multiple filters in series, but I don't think there is ever a valid justification for multiple filters in parallel (if you need more flow, get a bigger filter).

David

 

[BenjaminM]

Thankyou all for the responces.

I more than agree with you Zdas04 about having a single, larger filter. That is one of my long term goals for our process.

I will most likely go with Davsy's approach where the pressure is monitored on the pipe leading to the manifold, allowing us to switch between filters. In our case, the vessel being pumped out of is several feet above our filters. I believe we would still need a pressure sensor on the outlet side of the filters to negate the changing head pressure as the reactor is emptied.

 

[MortenA]

Since both filters feeds from the same point and deliver to the same point the dP must be the same (allthough flow is not).

You could measure dP by just measuring across one filter. Disregarding small losses it would be the same. However if that one filter is out of service....

Best regards

Morten