Reduced Scale Experiments on Fire Spread Involving Multiple Informal Settlement Dwellings

Round 1

Reviewer 1 Report

In this paper the results of an experimental campaign to study potential scaling between full scale (FSE) and reduced scale (RSE) tests considering structure to structure flame spread and fire growth are presented. The specific focus is to see if RSE experiments can be used to provide insight into FSE experiments of fire behavior in informal settlements. This goal is quite interesting and valuable, and a wealth of experimental results are presented from these small-scale tests, but I have some significant concerns given the reported results.

I encourage the authors to consider the key questions below and if/how unified conclusions can be made based on these RSE tests such that correlations could be developed that allow for the prediction of FSE behavior. In the end, I believe that would be the most meaningful outcome of this work - demonstrating how small-scale experiments can be used to provide guidance on the key features that impact full scale burning behavior, such that real decisions could be made regarding structure separation distance, spread rate for forecasting, the impact of future such as packing density or wind on fire behavior. It appears some of these key outcomes (capturing fire spread rates effectively in RSEs or capturing the sensitivity to key inputs such as wind or spacing) are not possible on the basis of the current RSE results, and that’s troubling. This could guide (and be informed by) not only safer fire behavior/reduced risk in informal settlements but also extends to other applications that involve large outdoor fires in the built environment.

Main concerns:

Concise summary of learned outcomes

The bulk of the text in this report reads as a summary of experiments, rather than a unified description of variations in results under different conditions, how equations for scaling or developed (i.e., how this scaling was determined beyond simply reducing dimensions to 1/4 or 1/5), the physics controlling behavior of interest, and ultimate trends that can be used to predict behavior at different scales. In the end, this feels more like a book report than a cohesive document where straightforward conclusions can be found regarding the factors of influence in fire development in FSE informal settlements and how this can be predicted from RSE results (It's especially difficult to see if quantitative predictions can be made based on this scaling).

In the conclusions, a summary of average flame spread rates measured in different tests is presented but there is no unifying description of how to interpret the small scale results to predict full scale behavior. Especially for a journal focused on scaling, it would be nice to show that either the 1/4 or 1/5 scale experiments allow for the prediction of the impact of each of the key features that are tested here (e.g., spacing, wind) on notable events such as time to ignition or spread rate. Then, validating this versus full scale experiments would show if these scaling laws hold and have further value. I cannot make such an interpretation or see such conclusions in this paper.

I believe one of the key issues that caused such deviations in measure dependencies between RSE and FSE experiments is that flame heat flux and flame exposure or flame length from an opening do not scale linearly as simply as the geometry has been scaled down from 1/4 to 1/5. As these are the forces driving flame spread from structure to structure, it suggests that the dynamics of fire growth at RSE in your tests do not match those at the larger scale, which calls into question the ability to extrapolate from results presented here. I'll summarize this with a quote from the current paper, " the fire spread rate was not captured accurately in RSEs". The follow-up to that quote that these experiments "provide a methodology to help further study..." Well, I'm not sure if that is the case if we’re not capturing the key feature of interest here. Is this the best platform/methodology to consider (as-is) or can more insight help get future experimental setup such that better correlation exists between scales?

Please define the controlling physics of the problem and present how scaling laws were developed and how that informed experimental design.

Throughout the paper, scaled times or distances or areas are presented but it is unclear to me how those are calculated. What are the scaling laws that are used, how and why were they developed ( specifically, what are the controlling physics here that they rely on ), and then how can these equations be used to calculate results presented in different tables? Without this, it is unclear to me where these different times come from. For example, in table four, scaling of time to flash over provides worse agreements with RSE results versus FSE results. My interpretation of this is that the scaling does not work and these RSE tests do not provide further insight to help guide understanding of full scale fire behavior

Full scale experiments showed a strong, critical dependence on wind speed, but many of the RSE results do not show that. That suggests to me that one of the key drivers of flame spread in the real system of interest are not captured here in the reduced scale tests. How can I interpret these results? What further insight can be gained from these RSE tests if this is the case?

How were previous modeling results leveraged to help with RSE experimental design & interpretation of results?

In the introduction, several previous studies from the authors are introduced noting the previous fire modeling is highly successful and can capture full scale experimental behavior. This poses some questions - 

1. if that fire modeling is so successful and can "effectively capture the full scale experiments", why is this reduced scale modeling needed? Wouldn't simulations be easier and cheaper to conduct?

2. Clarification should be provided to indicate what exactly these simulations can predict - temperature; rate of spread; fire size, peak or time development of that value; the impact of terrain or winds, packing density of neighboring sheds, and/or shed composition (wood/metal/cardboard)..?

3. Ultimately, some further clarity on what exactly was predicted (Temp, q”, HRR, spread rates…), how that was simulated, and if that allows for quantitative or qualitative prediction of what exact targets would be valuable. One thing that I noticed in reading some of those previous studies is that the reduced scale experiments show a factor of two difference in predicted heat flux, which I believe  is the key driving component of structure to structure spread (along with whether or not direct flame impingement occurs). If this heat transfer is so substantially off between scales, I wouldn’t expect spread rate to be similar (as noted in the conclusions of this report, “this is not accurately captures in RSEs”).

4. Additionally, one of those papers notes that "scaling may not be suitable with combustible cladding"; however, later throughout this paper it is noted that the dynamics are spread between different RSE structures is highly sensitive to ignition of cardboard liners inside of the corrugated steel outer shell. If this is the case, pointing to previous work showing that scaling can work but without combustible liners suggest that these results in the new study will have some troubles. Can you comment on the apparent disconnect here?

On page 4, a summary of some FSE experiments is presented but the results are not clearly stated. I am unable to determine what the key outcomes are - i.e., what are the impacts of cladding materials, spacing key features such as fire spread times or heat flux at a distance or ignite ability. Could correlations be developed that allow for quantitative predictions?

Minor comments:

Regarding information in the tables, how is the total ventilation factor defined in table 1? If it's A^H0.5, My simple and back of the envelope calculations do not appear to provide the same values listed in this table. For each of the calculated values across different tables, some clarity would help by having the equation it's used here along with the description of where they come from, as requested in the comment above.

Figure 7 notes a fifth plot, (e), that does not appear. I suspect similar ones may be missing elsewhere.

Typo, conclusions, line 815: “As noted above, tests [it] a wind tunnel would be…”

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report

Overall, this paper left a good impression on me with vast experiments and huge data. However, some issues should be addressed to help improve the quality.

i)   Pay attention to the Journal format due to disunity in the full text.

ii)  The style of paper is like a project report. The content is rich and there are no obvious
highlights, which poses problems for readers to find key information. You should
summarize the main message, give your own comments, and promote the idea you
want to express, rather than letting the reader enjoy exploring. Perhaps it is necessary
to develop the ability to understand what readers are searching for

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 3 Report

This paper develops a methodology for multi-dwelling ISD scaling such that large-scale spread phenomena can be captured. This paper carries out a series of RSEs to study the influence of (a) the number of dwellings, (b) the orientation of dwellings, windows, and door openings, (c) cladding material, (d) wind effects, (e) the distance between dwellings and (f) fuel load on the spread. The structure is clear and the results are as well. Future research developments were outlined. It can be accepted in its present form. 

Author Response

Thank-you for the feedback. No specific updates were required based on the comments. See updates made in red based on feedback from the other reviewers. 

Reviewer 4 Report

The article presents an experimental approach to study fire propagation in informal-settlement-dwellings (ISDs) and explores effects of various parameters such as material type, wind speeds and direction, distance between ISDs, etc., on fire spread rate. The authors propose reduced-scale-experiments (RSEs) to capture the most relevant trends of fire-spread seen in full-scale-experiments (FSEs). The overall paper is well written. Experimental setups are clearly defined and the results obtained are explained thoroughly. The few areas that require some attention are as follows

1. The comparison with FSEs is not thoroughly investigated. Usually to establish some connection between Reduced-scale and Full-scale burns; a correlation is established between typical fire spread parameters like, rate of spread, max temperature, etc., in terms of heat release rate (HRR). Since HRR is directly proportional to experiment scale (amount of fuel burnt), the measured parameters do not require further scaling.

2. Since the experiments were not conducted in a wind tunnel how are the wind conditions kept similar when comparing RSE results to FSE, except for may be under no-wind conditions? Are the wind speeds used in RSEs scaled as well? Is this the major factor causing discrepancy between RSE and FSE results?

3. Conclusion on line 797: "RSEs cannot currently be used for quantifying spread rates" is not completely correct. It can be true for the study and analysis conducted in this paper. Reduced scale models are often constructed as for predicting fire spread and typically established as a function of HRR, wind speed, fuel smoke point, etc. To the best of knowledge, these kind of correlations are widely used in successfully predicting the spread of wildland-fires.

4. Line 375, says that fire spread time for RSEs was lower than FSEs by about 45 to 90% for experiments without the timber clading. But the results shown in above table indicate opposite observation, i.e., fire spread time is FSEs is in general much higher than RSEs.

Author Response

Please see the attachment

Author Response File: Author Response.pdf