PSV Sizing for Fire Cases: Is a dynamic model worth the time?
Pressure Safety Valve Sizing
For many in our field, Pressure Safety Valve (PSV) sizing is considered to be a relatively simple task that can be performed in a matter of minutes by process engineers with enough experience. Many myths and misconceptions about this task have appeared over the years, as evidenced by a simple search on the internet and see hundreds of forums available on this topic. But if industry standards like API 521 exist and clearly state the steps to be followed, why is there so much misinformation out there? This communication aims to clarify a few points of clear concern relating to PSV sizing.
In principle PSV sizing should be straightforward:
- Estimate the fire heat duty using API 521 correlations or a more detailed heat transfer model. Both approaches depend on the vessel wetted area, environmental parameters, heat transfer coefficients, etc.
- Calculate the thermophysical properties of the fluid at relief temperature and pressure. In particular, the latent vaporization heat (λ_{v}) is required for sizing. This task is normally conducted using a commercial simulator.
- Follow the procedure described in API 520 to obtain the orifice size.
In simple terms, the flowrate to be relieved is determined as,
Equation 1
where Q̇_{fire} is the fire heat flow in kJ/h (Btu/lb), λ_{v }is the latent heat of vaporization in kJ/kg (Btu/lb), and ṁ_{relief }Is the relieving fluid flow rate in kg/h (lb/h). A conservative estimate of the orifice size must consider an adequate estimate of the mass flowrate at relief conditions, but still not too high that oversize the PSV. This estimate must consider a higher than usual fire heat flow and a minimum latent heat of vaporization. A reasonable question to ask is: how can we perform the calculation in a way that the risk of PSV oversizing is minimized? Around this question there are a few practices and misconceptions, for example:
- What latent heat of vaporization should be used to estimate the relief load? This question haunts process engineers. In a multi-component mixture, the latent heat of vaporization changes as the liquid is vaporized and the composition of the liquid in the vessel changes. This question becomes more pronounced as the mixture in question has a wide boiling point range. Should, for example, a minimum heat of vaporization be used? As the sizing procedure is often conducted with the aid of a commercial simulator, so, why not use tools in the simulator to make a more precise estimation of λ_{v}?
- Size for the “worse” case fire scenario: assume no insulation, drainage, or firefighting equipment available. This is in fact a common case in many smaller processing facilities.
- Assume there is a maximum heat transfer area: The area available to transfer heat to the fluid is clearly described by API 521 (wetted area). During a fire scenario the fluid vaporizes, so the wetted area decreases over time. Then it is reasonable to assume that the maximum area is the initial wetted area, thus the calculation always tends to overestimate the fire heat. If that is the case, why do some people recommend the use of transient wetted area estimation?
Case Study
A single vessel PSV sizing is considered here. A PSV must be installed for a horizontal vessel of known dimensions. A mixture of gas, water, and oil (38º API) enter the vessel that operates at 8ºC and 278.5 kPag. The PSV will have a set pressure of 1951 kPag. As per API 521, the fire case scenario considers an allowable overpressure of 2360.7 kPag (21% overpressure). The vessel is considered to be filled at 70% capacity for this scenario. We used Aspen HYSYS V10 to assist in our calculation. Initially, we employed the PSV sizing tool in the “Safety Analysis” section of Aspen HYSYS. This is a quick steady-state approach that closely follows the procedure described previously.
Estimate the fire heat according to the API 521 as (in USC units):
Equation 2
where F is an environmental factor (assumed here as 1 for a bare vessel with no insulation), and A_{wetted} is the vessel wetted area. Figure 1 presents the results sheet from the sizing procedure.
Figure 1: Results for a simple PSV sizing procedure.
For a wetted area of 477.6 ft^{2}, the resulting fire heat flow is 5.73 × 10^{6} kJ/kg using equation 2. Dividing this value by the estimated latent heat (equation 1), the estimated relieving flowrate is 3,683 kg/h, as shown in Figure 1. For this flow, the selected orifice is API 520 Size G (3.245 cm^{2} orifice area), which has a rated capacity of 4,891 kg/h, or 75.3% capacity use. The next smaller size (F, 1.980 cm^{2}) has a rated capacity of 2,984 kg/h. In this case, we can say there is a good error margin in the estimation. However, what if the estimated fire heat is too large and/or the latent heat is too large, thus the calculated flow rate is too small. In that case, the PSV would not be loaded at ~75% capacity but probably more (even as getting closer to 95%). The opposite scenario is also possible (oversizing the PSV), but less likely.
Figure 2: Variation of the latent heat of vaporization with a fraction of fluid vaporized.
For example, the latent heat of the relieving fluid is not constant but changes over time as the composition in the vessel evolves. As mentioned earlier, minimum latent heat is preferable for conservative sizing. Figure 2 shows the latent heat of vaporization of the fluid for different percentages of fluid vaporized (mimicking time evolution). An early estimate would give a large latent heat, leading to a small relieving flow that in consequence leads to an undersized PSV. For our case, we have chosen to work near the minimum latent heat (about 50% fluid vaporized, corresponding to 1,555 kJ/kg), so the estimate is conservative and yet covers a wide vaporization range.
Engineers must be cautioned to avoid using the latent heat of vaporization reported in the stream results (HYSYS or another simulator). This value assumes full vaporization of the liquid. If the mixture happens to have a small fraction of very heavy components, the reported latent heat can be quite low (a high temperature is required to boil a small fraction of the heavy components). For example, the presence of even 1 ppm of a heavy hydrocarbon will skew the latent heat of vaporization to an unreasonably lower value which in turn would result in an underprediction of the relief load.
The main takeaway from this exercise suggests that blind trust in the estimates of fire heat flow and latent heat may get us into an uncomfortable position. A good practice would be to double-check with some other methods. But what type of methods?
A common recommendation is to develop a dynamic simulation to improve the accuracy of the sizing calculations (this is even suggested in API 521). As this truly is a transient process, it is reasonable to consider that approach as superior to a steady-state model, however, dynamic models have usually the connotation of being unnecessarily time-consuming while not providing significant improvements. With modern commercial simulators, however, a dynamic model can be set up in few steps starting from a conventional PSV sizing scenario. For example, we have used the Dynamic Depressurization tool in Aspen HYSYS to check the performance of the selected PSV orifice, as obtained in the step before. The three main components for the simulation are the vessel dimensions, PSV specifications (orifice size, material, discharge coefficient), and heat transfer model.
For this report, we have considered three heat transfer models: (1) API 521 (equation 2), (2) an improved heat transfer model that considers convection of air and conduction effect through the vessel wall, and (3) the same heat transfer model as in (2) that also assumes a variable wetted area (as it varies over time). It is common to hear claims that assuming a constant wetted wall (instead of transient) may lead to significant errors during sizing.
Figure 3: Dynamic behaviour of a PSV during a fire scenario (case 1).
Figure 3 shows the dynamic behaviour for case (1). As heat is supplied to the vessel, pressure (and temperature) increases until reaching set pressure. This occurs at approximately 90 minutes, at which point the PSV starts to open. It was assumed that the valve is completely open at 10% overpressure (2,149 kPag). Pressure and relieving flow reach a maximum at about 140 minutes and then decrease over time. From this dynamic model, a PSV re-sizing can be performed using the peak-value conditions and flow rate. The peak flow is 3,689.5 kg/h (about 8,110 lb/h). Interestingly, the required flow is only 0.2% higher in the dynamic scenario compared to the steady-state approach. This kind of result seems to discourage the development of dynamic models for PSV sizing; however, some interesting results are obtained. For example, the steady-state approach results in a relief temperature of 182.5ºC vs 210ºC for the dynamic model (almost 30ºC difference). This difference can be crucial in some cases for the selection of materials and vendors, and even for the design of equipment and lines downstream of the PSV. Other than that, the dynamic model provides a check (probably peace of mind) to the process engineer.
Comparing the base case with the dynamic simulation we observe the estimated latent heat of vaporization is almost identical at some point in both cases, Figure 4. The latent heat increases monotonically until the fluid reaches the set pressure of the PSV. At this point, the latent heat in the dynamic model is almost identical to the steady-state calculation. However, after the valve starts to open the latent heat decreases as vapour starts to flow out from the vessel. At peak conditions, the latent heat is minimum, which as mentioned before is beneficial for sizing purposes. This means, that if the PSV sizing were only done based on a dynamic model, the relief flow would be slightly higher, thus providing an extra safety margin. Nonetheless, the selected PSV size would be the same with both approaches. The main comparison parameters selected for this study are provided in Table 1.
Figure 4: Evolution of the latent heat of vaporization over time for case 1. Opening of the PSV occurs at approximately 90 minutes.
Looking at the results of the case (2), we can observe the properties at relief conditions and relief flow are almost identical. This reassures that the correlation provided in API 521 to estimate the heat flow for a fire scenario provides enough accuracy and robustness. For case (3) we observed a very interesting behaviour although the final results are almost identical to the base dynamic case (case 1). In this case, the transient wetted area of the vessel is simulated. As the fluid is vaporized, the level in the vessel decreases (along with the wetted area). We do not see that behaviour in Figure 5 (at least initially). As the liquid temperature increases, it expands with not much formation of vapour.
Figure 5: Dynamic results for the transient wetted area model. Vessel wetted area (left) and heat flow (right).
Effectively, the wetted area increases (it's higher than at operating conditions), therefore the heat flow is higher than predicted with a constant wetted area model. This would mean a higher relief flow than produced by the steady-state model, however, after some time, the liquid vaporizes, and the level decreases start to decrease. This is less relief flow than predicted by the steady-state model. On average, these two effects are almost cancelled out and in the end, the results for both models look very similar. In fact, the net effect of considering a transient wetted area model is a higher relief flow, which leads to a higher required relief area. Of course, the extent of this behaviour cannot be generalized and must be verified on a case basis.
Table 1. Comparison of key results for the cases analyzed in this study.
Variable | Units | Steady State | Dynamics (1) | Dynamics Improved HT (2) | Transient Wetted Area (3) | Quasi Steady-State (4) |
T at relief | ºC | 182.5 | 210.03 | 209.9 | 209.2 | 194 |
λ_{v} | kJ/kg | 1,555 | 1,179 | 1,182 | 1,154 | 1,613 |
MW | kg/kmol | 35.67 | 30.31 | 30.26 | 31.13 | 38.12 |
Required Flow | kg/h | 3,683 | 3,689.5 | 3,594.4 | 3,722.1 | 3,761.9 |
PSV Capacity for G Orifice | % | 75.3% | 75.4% | 73.5% | 76.1% | 76.9% |
Another popular approach used by process engineers is considered in case (4). This method is based on Section 4.4.13.2.4.4 from API 521 (alternative methods). It consists of a quasi-steady-state simulation, in which first the vessel pressure is increased at constant volume, and then the temperature is increased in small time steps along with partial vapour return. This method is also known as multi-stage flash simulation with partial vapour return.
The continuous heating with an incremental heat input is used to determine the maximum relief load. To improve the precision of the calculation, smaller increments should be used. However, to limit the number of stages, smaller increments should be applied strategically near points where the system conditions are reaching flashing conditions for pure component phases.
We applied this method using the same heat flow from equation 2, and 5ºC temperature increments. Results are consistent with both steady-state and fully dynamic models, and in fact, the relief conditions obtained with this method lie between these cases. Even though the quasi-steady-state approach was widely used at some point (and still is even though being more tedious to set up), modern dynamic simulation capabilities and short computational times make this method almost obsolete. In fact, HYSYS has this method built into it which is known as semi-dynamic flash. A quick look into the results provided by this method in HYSYS show a slightly higher relief temperature (193.3ºC) and a required flow of 3,991 kg/h (81.6% capacity use for a G orifice size). As expected, these results being consistent with the other methods discussed here.
Figure 6: Quasi steady-state model approach developed in Aspen HYSYS. Each separator represents one temperature increment of 5ºC. A Total of 15 increments were simulated.
We can argue that the downsides of using dynamic models for simple engineering tasks like PSV sizing is are almost non-existent. The differences in time and effort between setting up a steady-state and a dynamic simulation are disappearing over the years. While it is true that in most cases the results between the two approaches are indistinguishable, for the few cases the sizing decision is not clear or too sensible to the model assumptions, the extra few minutes spent setting up a dynamic calculation well may be worth the effort.
Contact Process Ecology at [email protected] if you'd like to learn more.
References
- American Petroleum Institute Standards. API Standard 520 –Sizing, Selection, and Installation of Pressure-relieving Devices, Part I – Sizing and Selection. Ninth Edition. Washington D.C., 2014.
- American Petroleum Institute Standards. API Standard 521 –Pressure-relieving and Depressuring systems. Sixth Edition. Washington D.C., 2014.
- American Petroleum Institute Standards. API Standard 526 – Flanged Steel Pressure-relief Valves. Seventh Edition. Washington D.C., 2017.
- Firoozi, B. Wetted surface area calculation for fire-relief sizing in ASME pressure vessel. Hydrocarbon Processing, August 2015.
- Chen, G. Are you one of 99% engineers who size PSV fire case the wrong way?, 2015. Retrieved from:
- https://www.linkedin.com/pulse/you-one-99-engineers-who-size-psv-fire-case-wrong-way-guofu-chen-pes/
- Powers, C. Equations and Example Benchmark Calculation for Stepwise Fire Scenario Relief Loads. Aspen Technology Inc., 2017.
- Abouelhassan, M. Myths of Relief Analysis. Dynamic Relief – Safety by design.
- Available at https://www.dynamic-relief.com/
Interpretation Of Api 521 Fire Case And Removal Of Psv
I have a question regarding the use of API 521 3.15:
API 521 3.15 allows to eliminate the fire case for vessel located on structure above 7,6 m. The vessel is a propane receiver in a refrigeration system and has no other source for overpressure other than fire or thermal relief. The vessel can be isolated by automated valves from the PSVs that protects the system upstream from the overpressure during the shutdown of the system. In this case the vessel would not be protected against overpressure but there is also none of overpressure sources except pool fire and thermal expansion. In the shutdown case the vessel and adjacent piping that is isolated is only overpressure protected by system design (OPPSD)
Is it allowed or legal by rules of API 521 to eliminate the fire case for vessel higher than 7,6m, when the application of this 7.6m rule eliminates the only PSV(s) for the propane vessel (LPG!) and switches to OPPSD. In case of a fire there would no overpressure protection at all since the system pressure due to heat input even above 7,6 m (wetted area is currently 9m) would exceed the design and MAWP of the piping and equipment.
I believe the API 521 paragraph that eliminates the fire case for overpressure protection cannot be applied when it removes the only PSV(S) for the system, only when other relief case are exist. Since the vessel can be isolated the PSV covers the fire case based on the heat input for the elevation. Appreciate your thoughts in this matter.
Contact US
INTELLIGENT WORK FORUMS
FOR ENGINEERING PROFESSIONALS
Thanks. We have received your request and will respond promptly.
Log In
Come Join Us!
Are you an
Engineering professional?
Join Eng-Tips Forums!
- Talk With Other Members
- Be Notified Of Responses
To Your Posts - Keyword Search
- One-Click Access To Your
Favorite Forums - Automated Signatures
On Your Posts - Best Of All, It's Free!
Join Us!
*Eng-Tips's functionality depends on members receiving e-mail. By joining you are opting in to receive e-mail.
Posting Guidelines
Promoting, selling, recruiting, coursework and thesis posting is forbidden.
Students Click Here
Eng-Tips Posting Policies
Contact US
Fire Case for Small PlantFire Case for Small PlantProcessRookie(Chemical)(OP) Hi All, I'm doing the process design for a small new build facility, a near replica of an existing plant we have. The existing plant was designed by another party who just sized everything for the fire case without, I suspect, thinking too hard about what our credible worst case actually is. In the main processing part of the facility, the worst case liquid volumes that could credibly be released are very small (max 2m3 if the distillation column floods which it never has done to date) therefore I don't think it is credible that this type of release could sustain a fire under a vessel leading to a relief event. Does anyone know of any resource I can use to perhaps work out the fire area for the plant, or anything else I could use to assess whether or not the fire case is actually credible for each vessel? API 521 states any vessel at a height of less that 9m could be exposed to fire but I think that's based on large refineries where liquid volumes are significant throughout the plant. Cheers, PR The scientist describes what is; the engineer creates what never was. Red Flag SubmittedThank you for helping keep Eng-Tips Forums free from inappropriate posts. | ResourcesLow-Volume Rapid Injection Molding With 3D Printed Molds Learn methods and guidelines for using stereolithography (SLA) 3D printed molds in the injection molding process to lower costs and lead time. Discover how this hybrid manufacturing process enables on-demand mold fabrication to quickly produce small batches of thermoplastic parts. Download Now Design for Additive Manufacturing (DfAM)Examine how the principles of DfAM upend many of the long-standing rules around manufacturability - allowing engineers and designers to place a partâ€™s function at the center of their design considerations. Download Now Industry Perspective: Education and Metal 3D PrintingMetal 3D printing has rapidly emerged as a key technology in modern design and manufacturing, so itâ€™s critical educational institutions include it in their curricula to avoid leaving students at a disadvantage as they enter the workforce. Download Now Taking Control of Engineering DocumentsThis ebook covers tips for creating and managing workflows, security best practices and protection of intellectual property, Cloud vs. on-premise software solutions, CAD file management, compliance, and more. Download Now |
Fire can cause overpressure of storage or vessel. Either liquid vaporization of wetted vessel or vapor expansion of unwetted vessel due to heat input will increase the pressure. Heat input rate of fire exposure is not calculable from standard of heat transfer fundamental. Fortunately, OSHA regulations specify standards which are to be followed for particular material in and API also provides formulas for calculating heat input rate which are to be followed for particular condition of .
The following picture presents OSHA Venting Requirement of Fire Exposure to Storage Vessel.
For flammable and combustible liquid, heat input refers to NFPA 30 as follow.
Note :
1. Information in table applies to liquified components ; see original standard for nonliquified gases
2. CGA = Compressed Gas Association
NFPA = National Fire Protection Association
ANSI = American National Standard Association
3.Heat Input expressed in BTU/hr, with area in square feet.
4.See source documents for details of area calculations
My friend, if you want to check the correctness of value in above table, please refers to original reference, Guidelines for Pressure Relief and Effluent Handling System, AMERICAN INSTITUTE OF CHEMICAL ENGINEERS (AICHE), Table 3.3-1 (page 131 of 538). You can find that book in our library.
API STD 521 provides formulas for calculating heat input rate to process vessel containing liquid.
Q = 21.000 FA^0.82
Where adequate drainage and firefighting equipment do not exist, equation below should be used,
Q = 34,500 FA^0.82
Regarding a drainage, as long as the amount of liquid which is spilled out around the vessel is possibly decrease, use the first formula ( = heat flux 21000).
F is environment factor. The value F = 1 for bare vessel. Refer to API STD 521 Table 5 (page 17) for other than bare vessel.
A is wetted area of the vessel.
The value of A ^0.82 reflect that not 100% vessel area exposed by fire. It is true for large vessel, for small vessel, we can use assumption that 100 % vessel area is totally exposed by fire.
For example, for small bare vessel, the equation to be Q = 21.000 A.
My friend, please note, based on the formula, when 82% fraction area is applied the heat flux is 21.000 BTU/ft^1.64/hr or 34500 BTU//hr. And when 100% fraction area is applied, the unit will be BTU//hr. This understanding will useful in perform depressuring simulation using HYSYS.
The above formula is for pool fire case which has heat flux of 21.000 BTU/ft1.64/hr. Whereas, based on API STD 521, the heat flux for jet fire is 95.500 BTU/ft2/hr (average). From the unit of heat flux for jet fire, its show that 100% fraction area is used (for jet fire is localized heat flux).
Hope this drawing helps you get better understanding of pool fire and jet fire
And below is jet fire
My friend, in this posting we discuss about heat input rate of fire exposure for process vessel. The following are the important point,
1. For adequate drainage and fire fighting exist, use formula Q = 21.000 FA^0.82
2. For small vessel, use 100% fraction area is exposed by fire, therefore Q =21.000 FA. In my understanding, vessel with wetted area less than 200 ft2 can be considered small vessel (see NFPA 30)
3. For wetted vessel, overpressure caused by liquid evaporation, the relieving load capacity can be calculated as heat input divided by latent heat of vaporization. (W = Q/Hv).
4. Heat flux of jet fire is very high (95.500 BTU/ft2/hr average), and in localized area.
5. I suggest you to sizing PSV on fire case for pool fire case only.
6. The relieving temperature can be higher than the vessel collapse temperature. Vessel will collapse before increasing pressure reach PSV’s set pressure. Therefore, actually, PSV doesn’t provide sufficient protection. Depressuring is one of the additional protections against fire. (We will discuss about fire case depressuring both for pool and jet fire in other posting)
My friend, thank you for reading, please correct me if I'am Wrong.
Case api 521 fire
This one was too delicate, but the element. Opening your door, your legs sank into this element. A wave rushed in and licked them, taking you in. That first wave of delight and happiness ran through your body. You were not alone on this deserted shore.
Single Vessel BLOWDOWN Template (v10.0)The room resembles a dungeon of an ancient castle, adapted for torture. On the walls are hung various whips, handcuffs and collars, some kind of chains, strap-ons of different shapes and sizes, dildos and many other items. The purpose of which is not obvious to me. In the middle of the room there are two wooden pillars with iron rings at different heights, and between the pillars there is a block, into.
Now discussing:
- Artificial leaf screening roll
- Pmdg operations center
- Baking soda and adderall
- Fancy script tattoo
- Arcadia d3 reptile lamp
- She loves me instrumental
- Owen total drama
Grimaced a little. Also the bottle in the pope gives its own. She looked in the window.