I thought I might kick up a bit of discussion here in the refining forum with a recent challenge we had from a client's Fired Heater. I'll walk everyone through the problem, challenges with data, our solution process, and resources we used. I'd be thrilled to hear your ideas on this problem, and if you have questions please ask away!
The Problem
Client's fired heater was only able to hit 70% of its target duty and flow. They had to back off the heater to avoid "draft problems." The draft problems are described below:
Data Challenges
The Solution
We approached this study the way we always do: first, build a system model to match the design parameters. This helps to verify the original design and provides a starting point. Then, model in current operation conditions, and tune the model to depict the current system state. Lastly, with a tuned model, we can then apply changes and offer the client various routes forward.
Our first task was to get valid process data. We accomplished this by using DWSim, an open source process simulator. DWSim lets you input the distillation curve to get a stream of the crude. Then, you can run a "sensitivity analysis." In our case, we ran it at 3 pressures (20psia, 80, and 140), and temperature ranges from 340 to 880*F. Its important to have adequate bounds like this to capture film temperatures, which may be up to 50*F higher than the bulk temps. We made two data sets, one for the original crude, and one for the new crude.
The second task was to model the fired heater configurations. We did this with FiredHeaterPro. FHP is a web-based month-to-month fired heater simulator. I wrote it myself in fact! We used the following models to assess:
I'll spare you the gory details of each run, but cut to the chase on the general result. The 1980 model (#1) confirmed the original design, which was done without the aid of computers. The new convection section (#2) threw us for a loop. This model showed a tighter draft envelope (less draft, but still enough), but the client had told us that the new convection section made the heater run poorly ever since then. I asked the client if it had run ok for the first 2-3 years, and the confirmed that indeed it had run ok for a bit over 2 years before requiring turndown.
This was a nice confirmation of our model, which showed that the new convection section heater did have sufficient draft as long as the tubes were not heavily fouled. The reason that internal fouling causes draft problems is that it acts as insulation, and thus the operators must fire the burners harder in order to get the same duty. That means more fuel gas, more air, more flue gas, and thus more pressure drop and a loss of draft.
For the solutions, the client wanted us to consider cutting out rows, using bare tubes, and simply going back to the old convection section. They did not want to consider a blower to make up air, as that would void their permit. The bare tubes and row cutting would result in an extreme loss of efficiency, to the point that no duty would be gained even though the draft would be better. Rebuilding the old convection section would be prohibitively expensive.
Our simulations showed that in fact adding stack length was the most economical way to increase draft. A big caveat there is that you cannot just add stack length until the simulation hits its target. You must also consider the structural integrity of the stack as well as the increased foundation loading. See Annex H in API-560 for calc details.
In addition, the client is going to pig the heater tubes, and we will inspect the burner tips to ensure that all orifices are clean and within tolerance.
Further Resources:
DWSim
FiredHeaterPro
API-560
Direct-Fired Heaters by Newnham (text book)
John Zink Combustion Handbook (text book)
The Problem
Client's fired heater was only able to hit 70% of its target duty and flow. They had to back off the heater to avoid "draft problems." The draft problems are described below:
- Long lazy flame, whipping around the firebox.
- A couple of incidents where they over-fired and got a buildup of combustibles, resulting in a loud "pop" when they finally ignited.
- It had been over 7 years since the last pigging to clean the heater tubes.
- Client replaced their design convection section with a new section that had nearly double the area, going from continuous fins to serrated fins.
Data Challenges
- The client's data on their crude was pretty minimal, consisting of only a simple distillation curve.
- As-built drawings from the 1970s were of abysmal quality.
- There were no as-built drawings of the new convection section, and no engineering report or data on the section, just a brief email from the engineer.
- The heater's "bad behavior" had to be stitched together by our team from several conversations, and re-evaluated based on simulation results (more below on that).
- Some of the instrument data was clearly wrong, such as showing a hotter stack temp than firebox temp.
- Efficiency calc from stack instruments shows over 100% efficient, evidently breaking the laws of thermodynamics.
The Solution
We approached this study the way we always do: first, build a system model to match the design parameters. This helps to verify the original design and provides a starting point. Then, model in current operation conditions, and tune the model to depict the current system state. Lastly, with a tuned model, we can then apply changes and offer the client various routes forward.
Our first task was to get valid process data. We accomplished this by using DWSim, an open source process simulator. DWSim lets you input the distillation curve to get a stream of the crude. Then, you can run a "sensitivity analysis." In our case, we ran it at 3 pressures (20psia, 80, and 140), and temperature ranges from 340 to 880*F. Its important to have adequate bounds like this to capture film temperatures, which may be up to 50*F higher than the bulk temps. We made two data sets, one for the original crude, and one for the new crude.
The second task was to model the fired heater configurations. We did this with FiredHeaterPro. FHP is a web-based month-to-month fired heater simulator. I wrote it myself in fact! We used the following models to assess:
- Original 1980 as-built design (old crude, old fuel gas)
- Design with new 2020 convection section
- New convection section heavily fouled (present conditions)
- Old convection section heavily fouled (present conditions, crude, and new fuel gas)
- New convection section with modified stack.
- New convection section with 3 of 6 rows cut out.
- New convection section with bare tubes (client request).
I'll spare you the gory details of each run, but cut to the chase on the general result. The 1980 model (#1) confirmed the original design, which was done without the aid of computers. The new convection section (#2) threw us for a loop. This model showed a tighter draft envelope (less draft, but still enough), but the client had told us that the new convection section made the heater run poorly ever since then. I asked the client if it had run ok for the first 2-3 years, and the confirmed that indeed it had run ok for a bit over 2 years before requiring turndown.
This was a nice confirmation of our model, which showed that the new convection section heater did have sufficient draft as long as the tubes were not heavily fouled. The reason that internal fouling causes draft problems is that it acts as insulation, and thus the operators must fire the burners harder in order to get the same duty. That means more fuel gas, more air, more flue gas, and thus more pressure drop and a loss of draft.
For the solutions, the client wanted us to consider cutting out rows, using bare tubes, and simply going back to the old convection section. They did not want to consider a blower to make up air, as that would void their permit. The bare tubes and row cutting would result in an extreme loss of efficiency, to the point that no duty would be gained even though the draft would be better. Rebuilding the old convection section would be prohibitively expensive.
Our simulations showed that in fact adding stack length was the most economical way to increase draft. A big caveat there is that you cannot just add stack length until the simulation hits its target. You must also consider the structural integrity of the stack as well as the increased foundation loading. See Annex H in API-560 for calc details.
In addition, the client is going to pig the heater tubes, and we will inspect the burner tips to ensure that all orifices are clean and within tolerance.
Further Resources:
DWSim
FiredHeaterPro
API-560
Direct-Fired Heaters by Newnham (text book)
John Zink Combustion Handbook (text book)
