You're right:

The system performance curve comes from studying the friction characteristics of the system.

The pump curve usually comes from the manufacturer.

The intersection of two curves to determine the operating point.

You can find an excellent discussion on this topic in:
Pump Handbook
by Igor Karassik, William C. Krutzsch, Warren H.Fraser and Joseph P. Messina (Editors) 2nd Edition
Mc-Graw Hill Book Co.
ISBN 0-07-033302-5

I would like to add to electricpete's note that static head (positive = pump must elevate fluid, negative = gravity flow is possible, no elevation changes and closed system = 0 ft) must also be considered besides the friction losses in the system.
Again, the reference above contains excellent discussion of the topic.

PS: electricpete, congrats on your TipMaster nomination!!!

Two sites where is all is revealed. www.mcnallyinstitute.com & www.pump-zone.com (refer the pump handbook).

Congrats! for being Tip Master of the week Electricpete and you are right again asusual. Your posts in electrical too are very explanatory and humble. That is simply great.

By the way do you have joint responsibility in work for both electrical and mechanical works? (like engineering manager or something like that)

Regards,

Truth: Even the hardest of the problems will have atleast one simple solution. Mine may not be one.

Thanks guys.

It confirms my thought.

Regards to you all.

Do you have an idea of how I can draw the system curves for a scenario where the pump speed is constant and flow rates is throttled by a control valve at the outlet. How do I determine the pressure loss (or equivalent length) for the control valves if throttling to reduce the flow rates is done at 5 different steps. I believe I should have 6 system curves (including at zero throttling). Then one pump curve at constant speed.

Regards,
mucour

Basically, you draw the system and pump curve as described above.  The intersection point is the maximum flow through the system and is essentially where all the pump head is consumed by the line losses and downstream delivery pressure.

For a reduced flow, you can use the same curve.  The dP taken by the control valve is simply the difference between your pump head curve and the system resistance curve.  If the pump is putting out 100 psig at the new flow and the system resistance curves says you need 20 psi to get the new flow through the system, the control valve takes the difference or 80 psi.

Lets say that you have 0% throttling (this is 100% open valve), then 20, 40, 60, 80% throttling...following your description...100% throttling would be shut-off valve.
You may indeed draw 5 different curves (for 0/20/40/60/80% throtling), the 100% throttling would be the vertical axis (and your 6th curve).
Of course there are infinite number of system curves depending on the throttling.
The Resistance coefficient K for the control valve @ different positions is usually included in the manufacturer's data (flow vs. press drop, Cv, etc).
A very crude approximation  to K would be using Crane TP410 fig in page A-31 if you know the Cv of the valve at different throttling positions.
Also, in case you do not have valve data you could just plot the 0% throttle considering the press drop across the fully open valve (e.g. using Crane TP410 K-factor Table (starting on page A-26). This curve will intersect the pump curve at Qz.
Calculate 0.8Qz, 0.6Qz, 0.4Qz, 0.2Qz
Determine delta P from the pump curve for those flows.
Each curve could be now calculated from (e.g. for 0.8Qz)
delta P = (delta P@0.8Qz / (0.8Qz)^2)* Q^2

The difference between the delta P where each curve intersects the pump curve, and the curve for 0% throttling is the pressure drop required at the control valve at each of the p*Qz flows, where p = 0.8, 0.6, 0.4, 0.2

With the flow and pressure drop you may be able to determine a suitable control valve, but consider that the best range for operation of a control valve is between 60 to 80% of the full range... therefore when selecting the control valve you may want to consider only the flow and pressure drop for 20 and 40% throttling and then evaluate the other operating conditions... e.g. for a system where precise control is required at low flows you may even consider a shut-off valve for the "big valve", the one we just defined and a by-pass around that valve with a "small control valve" that allows control in the lower flow ranges.

Two notes:
1. Beware of the high flows... high flows may cause the pump to cavitate (if the impeller is too big for the design flowrate of the system) - i.e. your flow is to the right of the flow where the pump cavitates.
2. Beware of low flows, do not go to the left of the minimum flow for the pump as recommended by the pump manufacturer

Hope this is what you were looking for.
a.

Dear Abeltio,

I am studying your write-up.

I went to check my Crane TP410 and I can't find page A-31. Could you please tell me the edition of your reference.

Thanks also to the other contributors.

Regards,
mucour

I am using TP410 Twenty First Printing -1982
The graph title is Equivalents Of Resistance Coefficient K And FLow Coefficient Cv
In my copy Appendix A has 32 pages.
Hope this helps.
a.

The equation (an approximate equation) to convert Cv to K is:

Cv = 29.9*d^2/K^0.5

This is the equation used to generate the chart abeltio refers to on page A-31.

Right... note that:
d = inside diameter of pipe in inches
The equation TD2K refers to is also listed as equation 2-6 of the TP410 (page 2-10)
If it fits your system (low viscosity fluids) you can use:
Eq. 2-7... hope you have the paper.

Dear Albeltio,

Thanks.

I discovered that I am still using the 1979 edition.

Regards