Monotube Flash Boiler Design

[Toymaker]

Hi Toymaker, I see a tooling mark along the tube which I suppose comes from bending process.
Could be stress from bending tool and wall thinning (in that area seem to be very high contact forces and material push outwards) the reason for failure happening in that neutral fiber (or close to) of tube?
But other than that, bulging looks quite uniform.

Good catch NapierDeltic. The die marks you see on the tube are created by the rolling process and I believe are due to the groove in the die being over-size for the tube. I machined all the dies myself and sized them for use with 0.625" (5/8) after I decided to use 5/8 in my boiler, but before I actually had any of the tube in hand. In retrospect, I should have waited. Tubing sold in Thailand as 5/8" (16mm) is neither and actually measures between 0.59X to 0.60X resulting in my 0.625" dies not providing adequate side support to the smaller-than-expected tube. I didn't even notice the issue until after I had already rolled the first 15 meter pancake coil. Given the work that goes into making these boiler coils, I decided at that point to continue.

The dies for the 8mm tube shown below bending an aluminum tube do a splendid job, because the 8mm tube is actually 8mm. You can't see the 5/8" dies very well in the photo below,...but they are indeed the problem, or the smaller-than-should-be tube is the problem,...depends on your perspective.

Back to your question: I measured a few tube scraps I have laying around and determined the ruptured area actually occurred at a wall thickening area; the wall is 0.003" thicker than in most other areas around the tube. The tool mark area is the thickest section, measuring 0.006" thicker than other parts.

The first sample tube ruptured on the outermost bend where presumably the wall is at it's thinnest, so no surprise there. The wall thickness on the outermost bend on the sample tube we're discussing measures thinner than wall areas directly adjacent to the rupture. This is a bit of a head-scratcher for me,...lots of maybes and could-have-been's, but nothing definite.

 

[Toymaker]

Curious... in my experience (not a lot really!) of bending tubes, the tube hardly compresses the inside radius at all, but stretches all the material from the inside radius of the bending die to the outside. So for calculation purposes for Barlow's formula/Hoop stress calculations, I always take a worst case" that the outer diameter wall has been stretched from inner diameter material of known thickness, so the original tube wall thickness is reduced by the ratio of ID/OD. (Circumference ratio). The thinner value is then used in my designs. As the thinnest wall is on the outside diameter of the bend I expected a split failure to occur there, but now I think there are other factors affecting the tube, to generate the split on the side...
I guess, that Toymaker's bending dies (a wheel and die? - or set of dies and wheels) has some imperfection in the way they meet at the mid-line. Perhaps a slight mis-alignment? Or not quite the correct dies for the tube OD? - as the mid-line is quite pronounced.
Perhaps this is causing some trapping of material so it does not slide correctly in the dies/rolls and is causing some local thinning that should not occur? The apparent line/fold along the mid-line where rolls/dies meet will form a discontinuity in the wall of the tube, and a shape and work-hardened stiffness in the tube forming a stress concentration that has made the stresses focus where we did not expect the tube to fail.
As the tube is bent, the inside zone will be slightly compressed, and slightly work hardened in the process, and the outer half or more will be stretched, and work hardened proportionally to the amount of stretching. I suspect the split has occurred near to where there is a natural neutral axis between the compression and tension zones and consequential work hardening during bending. So stresses have been focused by the "fold-line" stress concentration into an area of lower work hardening from the bending process.
But that is purely my hypothetical guess!
Conclusion, improve the bender so the fold-line does not appear where dies and rolls meet. Or is it too late to make a new coil?
K2

At this point in the build & test process, is it really necessary to make a new coil? I've already hydrostatically tested this coil and proven that it's yield strength is twice what I need, which means the ultimate tensile strength will be even higher thereby providing an even larger safety margin. I'm going to go with what I have.

 

[Steamchick]

Fair enough.I was thinking the same, but because I am still an amateur, and very cautious, I would make new bending rolls to the correct size as re-roll the tube after annealing.
But I would do a trial piece first - a copy of the test piece that you burst - made exactly as you have done so far, then annealed, re-rolled in new rolls, and re-hydraulically hardened....
If it makes a significant improvement then you can decide if it is worth all the extra work. I was "trained" to be so cautious because the Company made 500,000 cars per year, so the cost of failure far outweighed the cost of remaking or reworking components.
Your's is a different situation.
Sorry I can't be useful this time
A difficult decision?
K2

 

[NapierDeltic]

Hi Toymaker,
Food for thought.
What about a test to fail on a straight sample of tube.
Of course, also here are a lot of maybes (vs bended tube interpretation), but at least would give a magnitude order to the reduction in strength.
Also measurements on tube thickness all around in section (fresh bended). Thinning in rupture area could be about same size as in the outermost tube generator. Outermost fiber should have also maximum stress of material from processing....guess.
It is strictly my opinion but if strength at 3 and 9 is about same size as at 12 o'clock (in section), I - for myself- would consider it acceptable, as long as I overload test my entire boiler.
Just wanted to highlight a possible explanation for failure location.
I learn a lot of helpful things from your work and I am grateful!
But I also understand you are digging uncharted territory and I understand meaning of word cautious.

 

[TimTaylor]

Any R123 that is burned as a result of leaks into the combustion chamber is a good outcome. The safety data sheet doesn't say that burning R123 decomposes into HCl and HF, rather it states when the temperature of R123 is raised above 250 C, (inside a sealed container) it will decompose into HCl & HF.

HCl and HF are both highly reactive acids which at the high temps inside the combustion chamber, will quickly react with other chemical products of combustion, thereby decomposing almost as quickly as they are produced.
Phosgene decomposes at 250 C, meaning the combustion gases will instantly decompose it.

Phosgene (COCl2) breaks down into carbon monoxide and chlorine gases, both of which are highly toxic.

Section 10 of the MSDS also states:
NORMALLY STABLE? (CONDITIONS TO AVOID):
The product is normally stable.
Avoid sources of ignition such as sparks, hot spots, welding flames and lighted cigarettes or unit heaters to prevent
formation of toxic and/or corrosive by-products. Avoid mixing with air or oxygen above atmospheric pressure.

 

[Toymaker]

Phosgene (COCl2) breaks down into carbon monoxide and chlorine gases, both of which are highly toxic.

Section 10 of the MSDS also states:
NORMALLY STABLE? (CONDITIONS TO AVOID):
The product is normally stable.
Avoid sources of ignition such as sparks, hot spots, welding flames and lighted cigarettes or unit heaters to prevent
formation of toxic and/or corrosive by-products. Avoid mixing with air or oxygen above atmospheric pressure.

Tim, I do appreciate your efforts to keep me safe, but I seriously doubt that even if all 4 Liters of R123 in the boiler tubes were to leak into the hot combustion gases that a large enough volume of CO would be generated as to present an actual health hazard,...the boiler will always be operated in an outdoor setting, leaving no chance for a lethal build up of CO. Chlorine gas is highly reactive and given the elevated temperatures of the hot exhaust gases that created it, it would most likely combine with any of the products of combustion it came into contact with, leaving no free chlorine molecules exiting the boiler.

The real risks in using R123 as a working fluid is if it decomposes inside the boiler coils, where the hydrochloric acid will react with any exposed aluminum metal, such as the AN-10 flare fittings I use. If a leak should occur outside the boiler container in one those flare fittings, then those nasty toxic gases would escape into the atmosphere. This can be mitigated by frequently checking the color of the R123 (clear is good, reddish-brown is bad) and monitoring the pressure & volume inside the working fluid reservoir.

 

[Steamchick]

Hi NapierD: Your statement: "Outermost fibre should have also maximum stress of material from processing...." is correct. But from the REAL results obtained by Toymaker, this is strongly indicative of either localised thinning or stress concentration on the "hoop stress" due to the line of distortion t the intersection of bending rolls that are not exactly the correct size for the tube. (As Toymaker explained). Personally, I favour an hypothesis that it is a combination of the 2 factors: And Toymaker is taking the approach that as he will only be working "at ~50% of the proven yield strength" then the tubes are adequately strong. This should be proven by his Hydraulic test on the whole assembly. If there is no permanent distortion anywhere on the tube assembly after he has withstood 2 x NWP for over 20 minutes, then he has tested to the limits required by UK standard, and if he tests at 2.8 x the NWP he will have proven the assembly to the ASME limit, that extrapolates for the loss of strength at the elevated temperature he will be using..
Toymaker, following your post#192: "I'm now confident the yield strength of my boiler tubes is at least 1000 psi.": would you consider doing a 30 minute test at 1400psi as your final Hydraulic pressure withstand test on the whole tube assembly, having worked up in steps to work-harden the assembly?
Following this, would you consider doing a STEAM test, using water as the fluid - to prove the pressure does no damage at elevated temperature? This test would need the burner (run up to maximum heat output) heating the full coil in its final assembly in the fire-box, with a pump of the correct capacity to simulate the cooling of the Freon, but with a Safety Pressure Relief valve set to 550psi to ensure the assembly does not exceed that pressure while containing boiling water.
This is to conduct a "regular" boiler test to prove the mechanical system. This should run for at least 20 minutes with the boiler feed pump simulating the temperatures you plan in service with Freon.
You can follow this by another test of your Electronic Controls to ensure you do not reach the breakdown temperature of Freon.
Just an idea, for you to consider?
K2

 

[Toymaker]

Hi NapierD: Your statement: "Outermost fibre should have also maximum stress of material from processing...." is correct. But from the REAL results obtained by Toymaker, this is strongly indicative of either localised thinning or stress concentration on the "hoop stress" due to the line of distortion t the intersection of bending rolls that are not exactly the correct size for the tube. (As Toymaker explained). Personally, I favour an hypothesis that it is a combination of the 2 factors: And Toymaker is taking the approach that as he will only be working "at ~50% of the proven yield strength" then the tubes are adequately strong. This should be proven by his Hydraulic test on the whole assembly.

A hydrostatic test up to 1000 psi on the entire monotube assembly was already successfully accomplished and reported in post#192; post#194 even shows a photo with the pressure guage in the foreground showing 1000 psi with the monotube assembly in the background.

If there is no permanent distortion anywhere on the tube assembly after he has withstood 2 x NWP for over 20 minutes, then he has tested to the limits required by UK standard,

With the exception of the 20 minute wait time at pressure, my coils seem to have passed the UK standard, but I made no attempt to time how long the monotube was at 1000 psi. Thinking back to how I performed the test, taking time to examine all brazed joints for any signs of ballooning and leaks, then taking time to position the hand pump such that the guage would be visible in a photo with the monotube in the background, then set up the shot for a good picture; time at pressure was probably somewhere close to 3, perhaps 4 minutes,...certainly no more than 5.

and if he tests at 2.8 x the NWP he will have proven the assembly to the ASME limit, that extrapolates for the loss of strength at the elevated temperature he will be using..
Toymaker, following your post#192: "I'm now confident the yield strength of my boiler tubes is at least 1000 psi.": would you consider doing a 30 minute test at 1400psi as your final Hydraulic pressure withstand test on the whole tube assembly, having worked up in steps to work-harden the assembly?

Before any further hydro testing is performed on the monotube coils, I would first do as I stated at the end of post#192: "If I can make the time needed for one more hydro test on a sample tube, I'm curious to try the 20 psi step procedure on a sample tube which would then be tested to burst pressure."

Before any of that happens, I first need to find and purchase a pressure guage with a 2000 psi max. The markings on my high pressure guage between 1000 and 2000 psi are in 200 psi increments which is not small enough to perform the 20 psi incremental steps needed for the working hardening procedure.

Following this, would you consider doing a STEAM test, using water as the fluid - to prove the pressure does no damage at elevated temperature?

I still plan to follow the testing process I stated back in post#185: "Part of my risk mitigation plan is to use water in the boiler for all initial tests and until I'm satisfied the ECU is working properly and will quickly shut down any potential over temps or pressures.

This test would need the burner (run up to maximum heat output) heating the full coil in its final assembly in the fire-box, with a pump of the correct capacity to simulate the cooling of the Freon, but with a Safety Pressure Relief valve set to 550psi to ensure the assembly does not exceed that pressure while containing boiling water.
This is to conduct a "regular" boiler test to prove the mechanical system. This should run for at least 20 minutes with the boiler feed pump simulating the temperatures you plan in service with Freon.
You can follow this by another test of your Electronic Controls to ensure you do not reach the breakdown temperature of Freon.
Just an idea, for you to consider?
K2

Getting the ECU to work (using water) is the next step, not the last step. A properly functioning ECU will make all subsequent testing both safer and easier. So, K2, I hope you're a patient individual because I still have much work to accomplish before I'm ready to boil any water at significant pressure.

 

[Steamchick]

Thanks Toymaker, I am simply impatient, because this is such an interesting project! You cover it so well, I like a daily "fix" - by reading your posts.
But don't worry if days/weeks go by with nothing, I am not planning on departing this life for a decade or 2 yet!
(Still want to extract lots of Money from the Pension fund! - At least 10 years before I get my due reward - profit - for my investment!).
K2

 

[Nick Hulme]

Perpetual motion ?? Nope, not at all, just normal physics and chemistry. R-123 simply takes far less enthalpy (heat) to make it boil compared to water.

Which means it has far less energy to transfer, you don't get anything for nothing here, your logic is indeed flawed and using something with a lower boiling point and a lower latent heat of vaporization has no advantage in this application, in fact there are significant disadvantages.

 

[Steamchick]

Hi, Just trying to get my head around this. The Freon has a significantly lower Enthalpy than water, hence carries proportionately less energy from "highest temperature and pressure" - to "lowest" in the turbine, at similar mass flow rates to Steam. (After the blades there is simply lost energy (escaped past insulation) and recovered energy (heat exchangers, etc.) ).
So.. trying to make sense (in my head) of "Another reason to use Freons in small power generators is their low volumetric expansion rate compared to steam. While steam turbines need 10 or 20 rows of blades to extract all the energy from steam, (because steam expands a lot), turbines using Freons need only 3 rows max to allow the Freon vapor to return to room temp.": I think you are saying that the turbine extracting a similar power to a steam version would need a far higher flow rate with Freon than with steam. Freon has a lower "joules per Kg", so the equivalent generator to a steam system needs more "Kg/Second", to achieve the similar "Joules/second". Am I on the right track?
K2

 

[Toymaker]

Hi, Just trying to get my head around this. The Freon has a significantly lower Enthalpy than water, hence carries proportionately less energy from "highest temperature and pressure" - to "lowest" in the turbine, at similar mass flow rates to Steam. (After the blades there is simply lost energy (escaped past insulation) and recovered energy (heat exchangers, etc.) ).
So.. trying to make sense (in my head) of "Another reason to use Freons in small power generators is their low volumetric expansion rate compared to steam. While steam turbines need 10 or 20 rows of blades to extract all the energy from steam, (because steam expands a lot), turbines using Freons need only 3 rows max to allow the Freon vapor to return to room temp.": I think you are saying that the turbine extracting a similar power to a steam version would need a far higher flow rate with Freon than with steam. Freon has a lower "joules per Kg", so the equivalent generator to a steam system needs more "Kg/Second", to achieve the similar "Joules/second". Am I on the right track?
K2

Brilliantly explained !! I would only add the obvious points; fewer rows of blades equates to smaller price, easier maintenance, and most of all smaller size and weight.

 

[Toymaker]

Which means it has far less energy to transfer, you don't get anything for nothing here, your logic is indeed flawed and using something with a lower boiling point and a lower latent heat of vaporization has no advantage in this application, in fact there are significant disadvantages.

Please read Steamchick's post #211, above.

ORC (Organic Rankine Cycle) power generators have been around for many years, I'm not inventing some new technology. As of 2017 there were just over 560 ORC power generators world wide, with a combined total capacity of 2749.1 MW. None of these generators use water (steam) as the working fluid,...nor do they all use R123, instead they use as a variety of Freons, solvents, etc. to best fit each application.

I encourage you to research ORC (aka. Organic Rankine Cycle). The attached pdf document should be a good starting point as it discusses the selection and performance of multiple working fluids.

 

[ajoeiam]

Brilliantly explained !! I would only add the obvious points; fewer rows of blades equates to smaller price, easier maintenance, and most of all smaller size and weight.

Hmm - - - no mention of the lower output from the system. (Grin - - - you're losing complexity, size, extremes of operation but you're also losing output - - - that all quite goes hand in hand.)

 

[Toymaker]

Hmm - - - no mention of the lower output from the system. (Grin - - - you're losing complexity, size, extremes of operation but you're also losing output - - - that all quite goes hand in hand.)

As with any turbine, power output is directly proportional to mass flow rate through the blade rows,...increasing the power output of even a single stage turbine is as simple as designing for higher mass flows. That means greater nozzle area and perhaps longer blades (depends on the design), but no increase in the number of blade rows nor additional complexity.

 

[Steamchick]

Perhaps I am wrong, but increased rows of blades , or stages of the turbine, is simply a way of increasing efficiency..... by extracting more of the power by the time the gas gets to the end of the final stage?
In my simple head.
1 stage can extract up to 50% from the gas's "available energy", 2 stages, 50% + 25%, = 75%,
3 stages, 50% + 25% + 12% = 87%....
Etc. except for the thermal, friction, and other system losses.
Or however it goes.... irrespective of the gas's enthalpy? Much more to do with momentum exchange between gas and blade? For a simple reaction turbine.....
But you may consider that blades are aerodynamically shaped to increase efficiency? I have used "50%" as a perfect case, where the system really has a much lower stage efficiency. Including intermediate reversing fixed blade stages, that re-direct the gas flow for the next stage, so in effect, we should star at 25%, not 50% in my simple example.
However, my simple arguement considers the Aerodynamic effects of modern turbine blades to increase the effectiveness of each blade by maybe 2% - 10% as we progress through the stages, making mode blade configurations a bit better in multi-stage turbines.
But as with most things, there is a point where the cost of extracting more energy holds a higher capital cost than the returns on fuel costs...
I also understand (VERY Simply) that nozzles are designed to give near supersonic maximum gas speeds at the initial gas injection point.... - maybe Mach 0.95? - to avoid supersonic shock waves damaging blades, while setting the initial gas stream as high as is practical?
I guess the speed of sound for the selected gas can be found from the web... and extrapolated to determine the nozzle size for the 500 psi injection pressure and flow. (Please don't ask me to do it... I'm not that clever!).

Another point: reading the treatise on gases supplied by Toymaker, #213, I read this as using heat input from a lower temperature source being the justification for using R123, rather than water, whic is suitable for higher temp erasure heat sourced systems. (Which is what we are studying).
Interesting!
K2

 

[Toymaker]

Perhaps I am wrong, but increased rows of blades , or stages of the turbine, is simply a way of increasing efficiency..... by extracting more of the power by the time the gas gets to the end of the final stage?
In my simple head.
1 stage can extract up to 50% from the gas's "available energy", 2 stages, 50% + 25%, = 75%,
3 stages, 50% + 25% + 12% = 87%....
Etc. except for the thermal, friction, and other system losses.
Or however it goes.... irrespective of the gas's enthalpy? Much more to do with momentum exchange between gas and blade? For a simple reaction turbine.....

You have the right ideas, just remember that all gases have a limit to how much they can be compressed (without cooling); that limit is known as the critical point. At the critical point both liquid and vapor will be present at the same time, and the vapor cannot be compressed further no matter how much additional pressure is applied. Since all vapors have a limit to how much they can be compressed, that necessarily means all vapors also have a max limit to how much they can expand. R123 expands from it's temperature-pressure critical point down to ambient temperature and pressure in just 3 rows of blades,...adding more blade rows would only create drag.

But you may consider that blades are aerodynamically shaped to increase efficiency? I have used "50%" as a perfect case, where the system really has a much lower stage efficiency. Including intermediate reversing fixed blade stages, that re-direct the gas flow for the next stage, so in effect, we should star at 25%, not 50% in my simple example.
However, my simple arguement considers the Aerodynamic effects of modern turbine blades to increase the effectiveness of each blade by maybe 2% - 10% as we progress through the stages, making mode blade configurations a bit better in multi-stage turbines.
But as with most things, there is a point where the cost of extracting more energy holds a higher capital cost than the returns on fuel costs...
I also understand (VERY Simply) that nozzles are designed to give near supersonic maximum gas speeds at the initial gas injection point.... - maybe Mach 0.95? - to avoid supersonic shock waves damaging blades, while setting the initial gas stream as high as is practical?
I guess the speed of sound for the selected gas can be found from the web... and extrapolated to determine the nozzle size for the 500 psi injection pressure and flow. (Please don't ask me to do it... I'm not that clever!).

Well designed Convergent-Divergent nozzles can accelerate gases up to Mach 2.0 as shown in this CFD diagram. Surprisingly, the Mach shock waves dont seem to damage the impulse blading.

Another point: reading the treatise on gases supplied by Toymaker, #213, I read this as using heat input from a lower temperature source being the justification for using R123, rather than water, whic is suitable for higher temp erasure heat sourced systems. (Which is what we are studying).
Interesting!
K2

Yes, at 200 C, water is indeed a good choice for a working fluid,...but only if you're willing to use 20+ blade rows

 

[HMEL]

You have the right ideas, just remember that all gases have a limit to how much they can be compressed (without cooling); that limit is known as the critical point. At the critical point both liquid and vapor will be present at the same time, and the vapor cannot be compressed further no matter how much additional pressure is applied. Since all vapors have a limit to how much they can be compressed, that necessarily means all vapors also have a max limit to how much they can expand. R123 expands from it's temperature-pressure critical point down to ambient temperature and pressure in just 3 rows of blades,...adding more blade rows would only create drag.



Well designed Convergent-Divergent nozzles can accelerate gases up to Mach 2.0 as shown in this CFD diagram. Surprisingly, the Mach shock waves dont seem to damage the impulse blading.

View attachment 149240



Yes, at 200 C, water is indeed a good choice for a working fluid,...but only if you're willing to use 20+ blade rows

Work that the turbine can do in a rankine cycle is a result of pressure volume work. This will depend on the pressure drop available from the turbine nozzle to the condenser. Secondly the blades and the allowable area for flow has to take into the possibility of choked flow which is dependent of the properties of the gas or vapor and its temperature. Its not a simple matter of how many blades are required. If choke flow occurs the mass flow rate is limited and thus work output is limited. So a design criteria is determine the mass flow, and calculate area required and thus the area of the nozzle block and blade passages to ensure sonic velocities are never reached for the mass flow conditions.

 

[Toymaker]

Work that the turbine can do in a rankine cycle is a result of pressure volume work. This will depend on the pressure drop available from the turbine nozzle to the condenser. Secondly the blades and the allowable area for flow has to take into the possibility of choked flow which is dependent of the properties of the gas or vapor and its temperature. Its not a simple matter of how many blades are required. If choke flow occurs the mass flow rate is limited and thus work output is limited. So a design criteria is determine the mass flow, and calculate area required and thus the area of the nozzle block and blade passages to ensure sonic velocities are never reached for the mass flow conditions.

I agree with everything you've said and would only add that all steam turbines are essentially a diffuser with blades inside which extract work from the flowing, expanding gasses. If you pull all the blades and shaft out of a steam turbine, you're left with a hollow conical shaped diffuser. The tapper of the turbine walls should be somewhere between 7 and 10 degrees; greater diverging angles guarantees turbulent flow. 7 degrees and less guarantees smooth subsonic flow.

Regardless of the working fluid, the area expansion (or divergence) must be limited to approximately 7 degrees of tapper.

 

[HMEL]

I agree with everything you've said and would only add that all steam turbines are essentially a diffuser with blades inside which extract work from the flowing, expanding gasses. If you pull all the blades and shaft out of a steam turbine, you're left with a hollow conical shaped diffuser. The tapper of the turbine walls should be somewhere between 7 and 10 degrees; greater diverging angles guarantees turbulent flow. 7 degrees and less guarantees smooth subsonic flow.

Regardless of the working fluid, the area expansion (or divergence) must be limited to approximately 7 degrees of tapper.

The sonic velocity V of a gas is determined by the constants R, K, g and the value of its temperature.
That is V= square root (gkRT). Note that this value increases rapidly with T. Seven degrees does not guarantee smooth subsonic flow. Turbulent flow is always a given because the Reynolds number are way above 2000. There are two types of turbine blades and each has a different way of causing a pressure drop. One is impluse and the other is reaction blade. The angles are derived from vector equations and the shape of the blades and the nozzle block is designed specifically for its application. This is a basic thermodynamic consideration. So you will need to consider the gas Temperature and then calculate the velocity through all the passages and if its over the sonic velocity choke flow will result.