Mutiple sources in a gravity-flow system
191 views
Skip to first unread message
Maxim F.
Jun 15, 2011, 12:45:12 PM6/15/11
to Epanet and Development
Hi everyone,
I am in the process of developing a framework / training material for
modeling gravity-flow water distribution systems using EPANET. I need
help modeling multiple spring intakes in an EPANET model... Here are
my findings so far:
The first challenge I encountered was to find out how to best model a
break-pressure tank component under EPANET. I now use pressure-
reducing valves (PRVs) to do so with a 0m pressure setting. This could
apply to collection chambers and sedimentation tanks as well.
The challenge I am now facing is to find out the best way to include
multiple spring intakes (sources) in the model with specific flow
yields for each. As an example, let's say you have a source of 0.2 LPS
located at EL=100 m. You have another source of 0.1 LPS at EL=50 m,
and you have a community and tapstands at EL=10m. The total water
demand for tapstands would be 0.3 LPS. The mainline would start from
the first source, and a branch would get the second source to the
mainline in a sedimentation tank or collection chamber at the junction
of the 2.
If you use reservoirs as sources, you have to find a way to force a
reservoir to have a specific inflow (which would be the yield for this
source), something like a flow control-valve (FCV) in EPANET. If you
don't do this, the model will calculate the inflow proportions coming
from the multiple reservoirs interacting together and you won't get
something that can be applied to the real life system.
The problem I have noticed with the FCVs is that I end up getting some
very messed up pressure and head results. My hypothesis is that in
order to reach the specified flow, EPANET iteratively adjusts the
pressure at the nodes upstream/downstream of the FCV. With multiple
reservoirs and FCVs, it gets even worse. Whereas I would simply want
to say: "There is 0.2 LPS coming from this reservoir, that's it !!".
Another option would be to use a node with a negative borehole. But I
don't feel comfortable with this option for the following reason. The
flow is OK, but model calculates a pressure for this borehole. It
seems very odd to me to have 20 m or 30 m pressures at a node that is
your source, whereas this should be at open air. Maybe another mix of
PRV would be possible ...
Please let me know if you have any idea/comments/suggestions that
could apply to this. EPANET files would be appreciated. If you have
experience modeling gravity-flow systems with EPANET, definitely let
me know as I am finding very little resources online !
Thanks in advance,
Maxim
I am in the process of developing a framework / training material for
modeling gravity-flow water distribution systems using EPANET. I need
help modeling multiple spring intakes in an EPANET model... Here are
my findings so far:
The first challenge I encountered was to find out how to best model a
break-pressure tank component under EPANET. I now use pressure-
reducing valves (PRVs) to do so with a 0m pressure setting. This could
apply to collection chambers and sedimentation tanks as well.
The challenge I am now facing is to find out the best way to include
multiple spring intakes (sources) in the model with specific flow
yields for each. As an example, let's say you have a source of 0.2 LPS
located at EL=100 m. You have another source of 0.1 LPS at EL=50 m,
and you have a community and tapstands at EL=10m. The total water
demand for tapstands would be 0.3 LPS. The mainline would start from
the first source, and a branch would get the second source to the
mainline in a sedimentation tank or collection chamber at the junction
of the 2.
If you use reservoirs as sources, you have to find a way to force a
reservoir to have a specific inflow (which would be the yield for this
source), something like a flow control-valve (FCV) in EPANET. If you
don't do this, the model will calculate the inflow proportions coming
from the multiple reservoirs interacting together and you won't get
something that can be applied to the real life system.
The problem I have noticed with the FCVs is that I end up getting some
very messed up pressure and head results. My hypothesis is that in
order to reach the specified flow, EPANET iteratively adjusts the
pressure at the nodes upstream/downstream of the FCV. With multiple
reservoirs and FCVs, it gets even worse. Whereas I would simply want
to say: "There is 0.2 LPS coming from this reservoir, that's it !!".
Another option would be to use a node with a negative borehole. But I
don't feel comfortable with this option for the following reason. The
flow is OK, but model calculates a pressure for this borehole. It
seems very odd to me to have 20 m or 30 m pressures at a node that is
your source, whereas this should be at open air. Maybe another mix of
PRV would be possible ...
Please let me know if you have any idea/comments/suggestions that
could apply to this. EPANET files would be appreciated. If you have
experience modeling gravity-flow systems with EPANET, definitely let
me know as I am finding very little resources online !
Thanks in advance,
Maxim
Santiago Arnalich
Jun 15, 2011, 1:33:12 PM6/15/11
to epa...@googlegroups.com
Hi Maxim,
In my humble opinion forcing flows is not a good option. That is what epanet is there for, you decide what consumption you want the system to deliver and then let the pipes take care of the flows.
In reality what happens is that friction through the two incoming branches for the flows on each branch have to be so that the water arrives with the same pressure. That can be calculated if the demand is contant. If it changes over time, then the pipes would take more whater from one spring or the other depending to make the headlosses match, and that implies that it may take more water from one spring or the other (then one of the springs will overflow), or you have another problem: Epanet is for fully filled pipes, but with such low flows pipes are more likely to fill only partially (not pressurized in parts of the way).
So:
In a general approach I would set the constant maximum demand and then change pipe diameters until you get the flows you want on the pipes. Since it is unlikely that a commercial diameter would do so, you would need to strangle the pipe at some point (with a gate valve and without epanet, by trial and error)
In the particular case you describe... why not forget about the issue and just place a BPT somewhere below the population? With that settup the BPT should be somewhere around 40m.
Regards,
Santiago
PD: I will contact you privatelly
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for relief work
In my humble opinion forcing flows is not a good option. That is what epanet is there for, you decide what consumption you want the system to deliver and then let the pipes take care of the flows.
In reality what happens is that friction through the two incoming branches for the flows on each branch have to be so that the water arrives with the same pressure. That can be calculated if the demand is contant. If it changes over time, then the pipes would take more whater from one spring or the other depending to make the headlosses match, and that implies that it may take more water from one spring or the other (then one of the springs will overflow), or you have another problem: Epanet is for fully filled pipes, but with such low flows pipes are more likely to fill only partially (not pressurized in parts of the way).
So:
In a general approach I would set the constant maximum demand and then change pipe diameters until you get the flows you want on the pipes. Since it is unlikely that a commercial diameter would do so, you would need to strangle the pipe at some point (with a gate valve and without epanet, by trial and error)
In the particular case you describe... why not forget about the issue and just place a BPT somewhere below the population? With that settup the BPT should be somewhere around 40m.
Regards,
Santiago
PD: I will contact you privatelly
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for relief work
--
www.arnalich.com
To send a message to the group, email: epa...@googlegroups.com
To unsubscribe, send a message to: epanetD+u...@googlegroups.com
Maxim F.
Jun 15, 2011, 3:01:11 PM6/15/11
to Epanet and Development
Hi Mr. Arnalich,
Thank you for your answer. I appreciate your general recommendation of
looking at the problem with a different perspective. However, I don't
feel comfortable in going a 100% with this and "letting the pipes take
care of the flows". If field surveys indicate that a given spring has
a maximum yield, it wouldn't be appropriate to have a higher flow
coming out of the source in the computer model ?
I will do some testing with "strangling" the appropriate pipes rather
than directly controlling the flow. I will come back soon with my
results.
Best regards,
Maxim
On Jun 15, 1:33 pm, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> <http://www.arnalich.com/>
Thank you for your answer. I appreciate your general recommendation of
looking at the problem with a different perspective. However, I don't
feel comfortable in going a 100% with this and "letting the pipes take
care of the flows". If field surveys indicate that a given spring has
a maximum yield, it wouldn't be appropriate to have a higher flow
coming out of the source in the computer model ?
I will do some testing with "strangling" the appropriate pipes rather
than directly controlling the flow. I will come back soon with my
results.
Best regards,
Maxim
On Jun 15, 1:33 pm, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
Santiago Arnalich
Jun 15, 2011, 3:20:04 PM6/15/11
to epa...@googlegroups.com
Hello Maxim,
In the example you provided I would set the demand for 0.3 l/s. That means that in total the flow would not exceed the combined flow of both.
Them I would work the diameters under until I get the right flows.
The thing is that if you want to use epanet, pipes have to be filled (pressurized). And a system like you describe is going to despressurize at some point almost certainly. If you follow the steps I described the pipe diameters would be correct since a partially filled pipe can transport more water than a fully filled one. Then the system will self-regulate as I described to have pressures match at the junction point. If too much water is demande from one source, the upper part of the pipe will empty.
Then the weight of the water (one of the componets of pressure in the Bernoulli equation), will be less and the pressure from the other source will "impose" its water until a new equilibrium is found. So if the demand is 0.3 l/s, the system will use both sources fully in any case, provided that the pipes can transport the yields needed.
Santiago
PS: You don't need to model everything with epanet. Most of the time breaking up chunks with simplifying rules is easier, because explaining how to build a whole model for every situation to beginners can be very challenging and prone to errors.
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for relief work
In the example you provided I would set the demand for 0.3 l/s. That means that in total the flow would not exceed the combined flow of both.
Them I would work the diameters under until I get the right flows.
The thing is that if you want to use epanet, pipes have to be filled (pressurized). And a system like you describe is going to despressurize at some point almost certainly. If you follow the steps I described the pipe diameters would be correct since a partially filled pipe can transport more water than a fully filled one. Then the system will self-regulate as I described to have pressures match at the junction point. If too much water is demande from one source, the upper part of the pipe will empty.
Then the weight of the water (one of the componets of pressure in the Bernoulli equation), will be less and the pressure from the other source will "impose" its water until a new equilibrium is found. So if the demand is 0.3 l/s, the system will use both sources fully in any case, provided that the pipes can transport the yields needed.
Santiago
PS: You don't need to model everything with epanet. Most of the time breaking up chunks with simplifying rules is easier, because explaining how to build a whole model for every situation to beginners can be very challenging and prone to errors.
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for relief work
Maxim F.
Jun 15, 2011, 4:22:16 PM6/15/11
to Epanet and Development
Santiago,
I understand that the total flow would now exceed the combined flow,
but my fear is that it's difficult to get the exact flows that you
would expect from the field measurements. But even those are a bit
inaccurate, depending on the season, the day, and etc ... Thanks for
the hydraulic explanation.
When working the diameters to get the right flows, you may end up
getting some pretty big headloss numbers on those pipes. Do you still
want to absolutely stay below the 10 m/km guideline or you can have
more considering you are trying to influence the flow path (and that
the velocities are OK) ?
Also, can you please provide some guidance on this: "Since it is
error ?
Breaking out into small models is a good piece of advice that I will
be putting to use too.
Regards,
Maxim
On Jun 15, 3:20 pm, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
I understand that the total flow would now exceed the combined flow,
but my fear is that it's difficult to get the exact flows that you
would expect from the field measurements. But even those are a bit
inaccurate, depending on the season, the day, and etc ... Thanks for
the hydraulic explanation.
When working the diameters to get the right flows, you may end up
getting some pretty big headloss numbers on those pipes. Do you still
want to absolutely stay below the 10 m/km guideline or you can have
more considering you are trying to influence the flow path (and that
the velocities are OK) ?
Also, can you please provide some guidance on this: "Since it is
unlikely that a commercial diameter would do so, you would need to
strangle the pipe at some point (with a gate valve and without epanet,
by trial and error) ". How should I proceed with this trial and
strangle the pipe at some point (with a gate valve and without epanet,
error ?
Breaking out into small models is a good piece of advice that I will
be putting to use too.
Regards,
Maxim
On Jun 15, 3:20 pm, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> Hello Maxim,
>
> In the example you provided I would set the demand for 0.3 l/s. That means
> that in total the flow would not exceed the combined flow of both.
>
> Them I would work the diameters under until I get the right flows.
>
> The thing is that if you want to use epanet, pipes have to be filled
> (pressurized). And a system like you describe is going to despressurize at
> some point almost certainly. If you follow the steps I described the pipe
> diameters would be correct since a partially filled pipe can transport more
> water than a fully filled one. Then the system will self-regulate as I
> described to have pressures match at the junction point. If too much water
> is demande from one source, the upper part of the pipe will empty.
>
> Then the weight of the water (one of the componets of pressure in the
> Bernoulli equation), will be less and the pressure from the other source
> will "impose" its water until a new equilibrium is found. So if the demand
> is 0.3 l/s, the system will use both sources fully in any case, provided
> that the pipes can transport the yields needed.
>
> Santiago
>
> PS: You don't need to model everything with epanet. Most of the time
> breaking up chunks with simplifying rules is easier, because explaining how
> to build a whole model for every situation to beginners can be very
> challenging and prone to errors.
>
>
> In the example you provided I would set the demand for 0.3 l/s. That means
> that in total the flow would not exceed the combined flow of both.
>
> Them I would work the diameters under until I get the right flows.
>
> The thing is that if you want to use epanet, pipes have to be filled
> (pressurized). And a system like you describe is going to despressurize at
> some point almost certainly. If you follow the steps I described the pipe
> diameters would be correct since a partially filled pipe can transport more
> water than a fully filled one. Then the system will self-regulate as I
> described to have pressures match at the junction point. If too much water
> is demande from one source, the upper part of the pipe will empty.
>
> Then the weight of the water (one of the componets of pressure in the
> Bernoulli equation), will be less and the pressure from the other source
> will "impose" its water until a new equilibrium is found. So if the demand
> is 0.3 l/s, the system will use both sources fully in any case, provided
> that the pipes can transport the yields needed.
>
> Santiago
>
> PS: You don't need to model everything with epanet. Most of the time
> breaking up chunks with simplifying rules is easier, because explaining how
> to build a whole model for every situation to beginners can be very
> challenging and prone to errors.
>
Santiago Arnalich
Jun 16, 2011, 2:31:27 AM6/16/11
to epa...@googlegroups.com
Hi Maxim,
I reply between lines
The system I described is a self-regulating mecanism, the only thing you need to worry about is that the pipes can carry the maximum yield of each spring. The system will take care of the rest.
No in my view. That is only an efficiency prescription, so that pipes are not loosing too much head that may need to be pumped or compensated with bigger diameters elsewhere. Many times you the headlosses are bigger that than. Then what you need to look are that velocities are not too big to cause damage (friction, water hammer etc) and that pressure is not excesive during periods of no consumption when water is not flowing so there is no friction --> excessive pressure
I reply between lines
On Wed, Jun 15, 2011 at 10:22 PM, Maxim F. <maxfr...@gmail.com> wrote:
Santiago,
I understand that the total flow would now exceed the combined flow,
but my fear is that it's difficult to get the exact flows that you
would expect from the field measurements. But even those are a bit
inaccurate, depending on the season, the day, and etc ... Thanks for
the hydraulic explanation.
The system I described is a self-regulating mecanism, the only thing you need to worry about is that the pipes can carry the maximum yield of each spring. The system will take care of the rest.
When working the diameters to get the right flows, you may end up
getting some pretty big headloss numbers on those pipes. Do you still
want to absolutely stay below the 10 m/km guideline or you can have
more considering you are trying to influence the flow path (and that
the velocities are OK) ?
No in my view. That is only an efficiency prescription, so that pipes are not loosing too much head that may need to be pumped or compensated with bigger diameters elsewhere. Many times you the headlosses are bigger that than. Then what you need to look are that velocities are not too big to cause damage (friction, water hammer etc) and that pressure is not excesive during periods of no consumption when water is not flowing so there is no friction --> excessive pressure
Also, can you please provide some guidance on this: "Since it is
unlikely that a commercial diameter would do so, you would need toby trial and error) ". How should I proceed with this trial and
strangle the pipe at some point (with a gate valve and without epanet,
error ?
In your model you can just put 1m of dummy pipe to account for the effects. In reality is very simple, you just install the valve and see what happens as you operate it. You get ana idea by opening a tap of your house and opening or closing the mains valve. Just trial and error.
One thing I didn't mention is that you also have to instal a check valve in the lower source to avoid that the higher spring discharges in the lower.
Hope it is helpful.
One thing I didn't mention is that you also have to instal a check valve in the lower source to avoid that the higher spring discharges in the lower.
Hope it is helpful.
Maxim F.
Jun 22, 2011, 4:58:20 PM6/22/11
to Epanet and Development
Hi Santiago,
A very simple question to come back to something you mentioned earlier
about the pressure during periods of no consumption.
If you remove the water demand at the taps to simulation condition
with no flow, you get a "System Unbalanced" warning message. However,
the pressure is still computed at the different nodes. Is this
pressure information accurate even though we got the warning message?
Also, in real life I guess the water would flow from the spring up to
the first overflow. Is this first overflow at the spring intake or
only at sedimentation tanks/break-pressure tanks.
Maxim
On Jun 16, 2:31 am, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> Hi Maxim,
>
> I reply between lines
> ...
>
> read more »
A very simple question to come back to something you mentioned earlier
about the pressure during periods of no consumption.
If you remove the water demand at the taps to simulation condition
with no flow, you get a "System Unbalanced" warning message. However,
the pressure is still computed at the different nodes. Is this
pressure information accurate even though we got the warning message?
Also, in real life I guess the water would flow from the spring up to
the first overflow. Is this first overflow at the spring intake or
only at sedimentation tanks/break-pressure tanks.
Maxim
On Jun 16, 2:31 am, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> Hi Maxim,
>
> I reply between lines
>
> read more »
Santiago Arnalich
Jun 23, 2011, 2:01:31 AM6/23/11
to epa...@googlegroups.com
Hi maxim,
If you are getting the error but still get pressure values, you are probably reading those prior to changing the demand. That is a variant of one of the most common errors, to make changes in the model and forget to recalculate the network.
The System unbalance error means that Epanet couldn't solve the equations in the number of trials that is given. The solving of a system is an iterative process where on each round it gets closer to the result.
Go to Data/Options/Hydraulics in the browser and change the field Maximum trials to a bigger value. That may solve the problem.
Nonconvergence can be a number of other things, like very low speeds with Darcy Weisbach equations and the computing of f value (Are you using it?)
Attach the system and we can take a look at it
Santiago Arnalich
Coordinador
+34 950 16 72 17
If you are getting the error but still get pressure values, you are probably reading those prior to changing the demand. That is a variant of one of the most common errors, to make changes in the model and forget to recalculate the network.
The System unbalance error means that Epanet couldn't solve the equations in the number of trials that is given. The solving of a system is an iterative process where on each round it gets closer to the result.
Go to Data/Options/Hydraulics in the browser and change the field Maximum trials to a bigger value. That may solve the problem.
Nonconvergence can be a number of other things, like very low speeds with Darcy Weisbach equations and the computing of f value (Are you using it?)
Attach the system and we can take a look at it
Santiago Arnalich
Coordinador
+34 950 16 72 17
Maxim F.
Jun 23, 2011, 10:06:14 AM6/23/11
to Epanet and Development
Hi Santiago,
Actually, the pressures seem to be part of new calculated data. The
flow is 0.0 in the pipes, but there is pressure at the nodes. I did
the calculations and the pressure is directly considering the
elevation difference. Sometimes it is cumulative, sometimes not
depending if there is a PRV in function.
Maxim
PS. I don't seem to have the attachment option in this google group, I
noticed it last week. I can find nowhere a way to attach a file with
my message.
On Jun 23, 2:01 am, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> ...
>
> read more »
Actually, the pressures seem to be part of new calculated data. The
flow is 0.0 in the pipes, but there is pressure at the nodes. I did
the calculations and the pressure is directly considering the
elevation difference. Sometimes it is cumulative, sometimes not
depending if there is a PRV in function.
Maxim
PS. I don't seem to have the attachment option in this google group, I
noticed it last week. I can find nowhere a way to attach a file with
my message.
On Jun 23, 2:01 am, Santiago Arnalich <coordinac...@arnalich.com>
wrote:
> Hi maxim,
>
> If you are getting the error but still get pressure values, you are probably
> reading those prior to changing the demand. That is a variant of one of the
> most common errors, to make changes in the model and forget to recalculate
> the network.
>
> The System unbalance error means that Epanet couldn't solve the equations in
> the number of trials that is given. The solving of a system is an iterative
> process where on each round it gets closer to the result.
>
> Go to Data/Options/Hydraulics in the browser and change the field Maximum
> trials to a bigger value. That may solve the problem.
>
> Nonconvergence can be a number of other things, like very low speeds with
> Darcy Weisbach equations and the computing of f value (Are you using it?)
>
> Attach the system and we can take a look at it
>
>
> If you are getting the error but still get pressure values, you are probably
> reading those prior to changing the demand. That is a variant of one of the
> most common errors, to make changes in the model and forget to recalculate
> the network.
>
> The System unbalance error means that Epanet couldn't solve the equations in
> the number of trials that is given. The solving of a system is an iterative
> process where on each round it gets closer to the result.
>
> Go to Data/Options/Hydraulics in the browser and change the field Maximum
> trials to a bigger value. That may solve the problem.
>
> Nonconvergence can be a number of other things, like very low speeds with
> Darcy Weisbach equations and the computing of f value (Are you using it?)
>
> Attach the system and we can take a look at it
>
> <http://www.arnalich.com/>
> Santiago Arnalich
> Coordinador
> +34 950 16 72 17
> +34 671700686
> skype: sarnalich
> ...www.arnalich.com
> Santiago Arnalich
> Coordinador
> +34 950 16 72 17
> +34 671700686
> skype: sarnalich
> ...www.arnalich.com
>
> read more »
Santiago Arnalich
Jun 23, 2011, 10:36:20 AM6/23/11
to epa...@googlegroups.com
I attach a .net file for a test.
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for Development
Santiago Arnalich
Coordinador
+34 950 16 72 17
+34 671700686
skype: sarnalich
...
www.arnalich.com
Consultancy & training for Development
> ...
>
> read more »
Reply all
Reply to author
Forward
0 new messages