Pressure spike damage to heat exchanger 1

[CheemaJ]

Hi,

We're running a hot water ring main and have experienced some issues with pressure spikes bursting gaskets on a plate and frame heat exchanger.

Water is pumped from a 20,000L tank via. centrifugal pumps that are powered by variable speed drives. There is a 100m pipe run before the water is heated in a plate heat exchanger by low pressure steam. The loop then goes off into the plant where there are individual takeoffs for process users. Air actuated spring return ball valves are used to control flow to each user as on or off (the valves are not modulating). At the end of the loop there is a pressure transmitter and a backpressure valve before returning to the tank. The pump speed is controlled off a set point on the pressure transmitter. So as users start to draw hot water, the line pressure drops and the pump ramps up to try and achieve the set point. The pump then slows back down.

When a valve closes, there is a big spike in pressure upstream - which has lead to gaskets leaking on the plate heat exchanger.

I've considered a few options to mitigate this and was wondering if anyone had any thoughts or suggestions on these:

1) Restrict the solenoid exhaust from the spring return ball valves, to dampen the pressure spike. The problem with this is that there are flow meters and additional valves downstream of the takeoff valve. So there'd have to be adjustments made to ensure the flow totaliser doesn't overshoot and also restrict exhausts of other downstream valves. This isn't very practical.
2) Install an expansion vessel to absorb the spike in pressure. I am guessing the best place for this would be on the water inlet to the heat exchanger. I have not sized an expansion vessel before, is there any resources anyone could recommend on sizing for this type of application?
3) Install a pressure relief valve next to the heat exchanger. I would prefer not to do this as it would mean wasting energy/water.

Thanks in advance

 

[LittleInch]

A schematic / sketch always helps and some data such as

Number of offtakes?
Size of the header?
flow velocity in the header?
Speed of closure at the moment (seconds).

How does this control loop work exactly?
If you've got a back pressure valve this implies that it tries to maintain a fixed back pressure of ?. Hence what is the pressure transmitter set at?
If say you've set it at 3 bar t maintain, as soon as the pressure goes above 3 bar if opens up? so what is the purpose of the VFD's? Are you setting the pressure TX to lower than the back pressure set point?

Normally in surge / spike you try to either slow the valve closure down, reduce velocity or introduce somewhere for the pressure wave to go.

I would put this on the discharge of the water heater myself. You're trying to stop the spike going back into the HX, so if the shock waves are coming from the on off valves then you want it between them and the HX.

How big?
Probably about 5% of the total water volume in the piping system with the vessel abut 30-50% full.
Or get a larger header.
But think about the control system as well - a fixed speed pump could be a better solution

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[curtis2004]

Hi,

I agree with LittleInch that some numbers would help a lot with schematics, even hand sketch.

Can you also give us detail of hot water consumers? What kind of process is it? It sounds that water is not returned back to heater and replaced with 100% make-up.
I think, the problem is Air actuated spring returns in valves. Is there any delay programmed for closure or spring shuts off it quick?

Thanks,
Curtis

 

[EdStainless]

So if this is a ring main then there is always return flow back to the HX correct?
Even when all of the branch demands are being used you still have excess flow back to the supply?
If this is the case then I am having trouble seeing how you can generate large enough spikes to cause issues.
However if this is just a header and you are just using a VFD for pressure control then you need to make sure that your valves cannot act faster than your VFD.
There should be at least a return from the far end of your header to the supply.
A sketch of the loop would be helpful. I am used to full ring headers in 6" or 8" with continuous bleed back to the supply via a 2" line with some restriction in it. The pump runs more, but cycles less.

= = = = = = = = = = = = = = = = = = = =
P.E. Metallurgy, consulting work welcomed

 

[CheemaJ]

Hi all, thanks for the prompt replies.

I've attached a snippet of the P&ID for this which should hopefully address a few of the questions you had asked.

The maximum velocity in the 2" header is 2.2 m/s. It is much lower than this when hot water is not being called for.

The valve closure speed is fast and though I've not timed it I'd say it would probably be a second, maybe one and a half seconds.

The back pressure valve is intended to create back pressure for the takeoffs whilst still allowing flow to pass by it, so the contents of the tank can be continuously heated by the heat exchanger. The idea (rightly or wrongly!) was that the hot water pumps would run at low speed for majority of time for energy saving purposes. The pressure transmitter is used to detect when a user calls for hot water, i.e. the pressure falls. The pump speed is then increased to deliver more flow. The valve will shut before the pump has time to slow down. I was hoping that the back pressure regulator (which is effectively a relief valve) would relieve the pressure spike but this hasn't been the case unfortunately.

The water is serving batch mixing processes and clean in place systems. There is no delay programmed for the closing of the valves.

EdStainless - maybe ring main was not the correct term for this and it is as you say a header.

It does sound as though we need to work to slow down the closing of the valves, though I will also investigate the installation of a surge vessel as the volume of the pipework is not significant (~ 1,200L).

 

[LittleInch]

Your other option is to re-configure the flow and pressure design here.

It is actually a bit of a myth that VFDs work to reduce power consumption in this sort of situation.

You haven't provided us with the set points and control points of the back pressure valve and the pump set point.

Apologies if this is simplistic.
A simple back pressure valve will try to maintain a fixed back pressure by varying the opening of the valve. As pressure increases above it's set point it opens more, as pressure goes below the set point it closes more to try and maintain the fixed set point. I'm at a bit of a loss as to how a pressure controller on a VFD driven pump works with this scenario as all I can see is two systems fighting each other.

I suspect what might be happening is that if the pressure set point for the VFD is below the set point for the valve, then as the pressure drops as a user opens the valve, the valve closes completely to try and maintain the fixed back pressure and the VFD then controls on this lower set point and then when a single user turns off the valve the pressure spike / surge has no where to go.

I think you may be much better served with a valve which allows a fixed flow to maintain the temperature in the tank and a fixed speed pump. In low flow cases the pump will consume a lot less power than at full flow. Just because the motor says say 20kW, doesn't mean it consumes 20kW all the time. Power consumption is flow dependant and I think this is one case where a VFD doesn't deliver the savings you think it is. Add in the VFD losses and you might be burning a lot of power.

The advantage here is that with constant flow the impact of the sudden closure should be greatly reduced and the lack of fighting between the two control systems will avoid sudden surges in pump speed.

Or you could change the back pressure valve to a constant flow valve and still allow the VFD to work, but some percent of the flow (10?, 15?) still goes back into the tank.

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[CheemaJ]

Apologies I forgot to add set points. During commissioning we adjusted the backpressure regulator spring to it's lowest setting. We ran the water pump and found that we had sufficient return flow to the hot water tank when the pump was running at around 30 hZ. The pressure transmitter was reading 1 barg at this speed. So the pumps were set to control the pressure to 1 barg.

I agree that when a user takes off water, the valve will close to maintain upstream pressure whilst the pump ramps up at the same time. So you're right they are competing with one another with no real benefit.

I've had an idea that may help realise some energy savings, what are your thoughts:

[ol 1]
[li]Remove the backpressure valve and replace it for an actuated ball valve with a hole drilled through the center of the ball[/li]
[li]When the pumps are running and no water is called for the pumps are run at a fixed speed (no pressure control). The ball valve is fully open, thus allowing good flow at a relatively low pump speed (lower energy consumption)[/li]
[li]When water is called for, the ball valve is closed (to create backpressure) and pressure control is enabled - thus allowing the pump to ramp up and deliver higher flow when needed[/li]
[/ol]

When the user valve is closed and the pump starts to slow down to fixed speed, would there still be the issue of pressure surge though? Maybe this is un-necessarily complicating things! So, to run the pumps at 100% fixed speed, in theory I could replace the back pressure valve with an orifice plate?

Thanks again for your help

 

[LittleInch]

There are many options - depends how sophisticated you want to get.

Setting both controls at the same pressure isn't a good plan in my opinion.

As you say above, you need a reasonable flow to keep the main tank warm / hot. I would change that back pressure valve to be a constant flow valve. I've not used these, but they seem to be pretty good constant flow with different back pressure valves.

https://www.braecosales.com.au/products/control-valve/maric-flow-control-valve/

Then you can either got to 100% on speed or continue to vary the speed depending on pressure, but set higher than 1 barg.

Or as you say just add an orifice plate and control to some higher pressure to have a fixed flow through the system to the tank.

Or make the valve a control valve, measure the flow and control to a flow rate.

But putting high pressure spikes through your HX is going to kill the HX sooner rather than later....

Remember - More details = better answers
Also: If you get a response it's polite to respond to it.

 

[georgeverghese]

Another option is to drill a 3mm hole through the one of the plates on the check valve at each pump discharge - the pressure spike would dissipate though this hole. Else replace the check valves with damped check valves. Else, if you dont mind losing a a trickle of water, crack open the bleed valve upstream of the steam heater all the time; that should dissipate the pressure spike too. Another option is to install a small bleed/bypass across the backpressure control valve and have it cracked open all the time. The flow through this small bypass should be lower than the pump throughput when at low speed.

 

[TBP]

I don't see a relief valve on the hot water side of the HX. This arrangement is EXACTLY like a hot water boiler installation, with the HX replacing the boiler.

I wonder if the steam valve doesn't leak-by a little bit, and at periods of low demand you're having pressure due to thermal expansion blow the gaskets in your HX.

No matter what, install a relief valve.