A Control-Oriented Model for Polymer-Dispersed Liquid Crystal Films as an Actuator for Natural Light Control

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

This paper describes continuous control of photo-synthetically active radiation through PDLC film. The suggestion of novel control-oriented model and their successful application to light intensity modification can be appropriately done for the publication of the journal. However, I have several questions to understand the research more clearly and to improve the reader’s understanding as follows:

1. In Fig. 1(a), the LC director at light scattered state should be more randomly oriented structure such as bipolar or radial. Please check the other references.
2. The reason for using PI control instead of PID or PD? Could you briefly comment on the choice?
3. In Fig. 6, the reason of two “R”s in the figure? (though there is no potentiometer)
4. In Fig. 11, after fitting, what is the use of T(v)? Since the key result in this study is continuous control of transmittance in terms of PWM control voltage applied to PDLC, Fig. 11 is a very important one. But after getting T(v) equation, it is not clear how it is implemented to the simulation after that. (i.e., for the simulation in Fig. 12?) Please add more explanation about this. (other than in page 9, line 264-265)
5. In Fig 12, the y axis is better to define as ‘normalized PAR’ instead of 'normalized output’.
6. The reason for normalization in Fig. 12?
7. In the step-down process of Fig. 12, the reason of using ‘negative value’.
8. Please add more discussion about the meaning of Fig. 12.
9. The simulation process in Fig. 12?
10. In the disturbance rejection, I suggest the use of other word than “setpoint” to make it more clear meaning. (i.e., minimum disturbance level)
11. The reason for suggesting setpoint value (21.5). It is clearly smaller than the input signal’s disturbance (30.0~37.3), but if possible, add more discussion for the setpoint value. (though setpoint tracking is also possible)
12. Since the main application of PDLC control-model is plant growth, if possible, please give an example or add a discussion about applicable plant based on the PAR in this study.

Author Response

The authors thank the reviewer for the insightful comments.  We endeavored to address all of them in the revised draft and the discussion below:

1. In Fig. 1(a), the LC director at light scattered state should be more randomly oriented structure such as bipolar or radial. Please check the other references.

Adjusted the figure. Thank you for the suggestion.

1. The reason for using PI control instead of PID or PD? Could you briefly comment on the choice?

Additional explanation was added to the text in line 109 and following.

1. In Fig. 6, the reason of two “R”s in the figure? (though there is no potentiometer)

This was done for expedience as this value of resistor was readily available and we wanted to report accurately what we did. Therefore, two resistors are shown in the figure.

1. In Fig. 11, after fitting, what is the use of T(v)? Since the key result in this study is continuous control of transmittance in terms of PWM control voltage applied to PDLC, Fig. 11 is a very important one. But after getting T(v) equation, it is not clear how it is implemented to the simulation after that. (i.e., for the simulation in Fig. 12?) Please add more explanation about this. (other than in page 9, line 264-265)

Additional clarification was made in the text.

1. In Fig 12, the y axis is better to define as ‘normalized PAR’ instead of 'normalized output’.

We used “output” instead of “PAR”, as by normalizing it becomes unitless as required for the context of the model in Figure 3. The gain is unit less therefore the fit function must be.

1. The reason for normalization in Fig. 12?

Thank you for the comment we added more explanation in the text.

1. In the step-down process of Fig. 12, the reason of using ‘negative value’.

If we went from one to zero, we would expect the same results. As we do not assume exactly the same behavior this is to emphasize on the fact that it ended up being the same.

1. Please add more discussion about the meaning of Fig. 12.

Thank you for the comment. More explanation was added in the corresponding section.

1. The simulation process in Fig. 12?

Thank you for the insightful comment. We added the specification that we used MATLAB lsim to the paper.

1. In the disturbance rejection, I suggest the use of other word than “setpoint” to make it more clear meaning. (i.e., minimum disturbance level)

Thank you for the comment the reasoning for calling setpoint is that it represents r in Figure. This is the value set for the control to keep for the disturbance rejection.

1. The reason for suggesting setpoint value (21.5). It is clearly smaller than the input signal’s disturbance (30.0~37.3), but if possible, add more discussion for the setpoint value. (though setpoint tracking is also possible)

Thank you for pointing this out. As the PDLC reduces light by at least 25% as shown in Figure 11 the setpoint must be significantly lower than the external light.

1. Since the main application of PDLC control-model is plant growth, if possible, please give an example or add a discussion about applicable plant based on the PAR in this study.

Thank you for the comment. A Reference 2 has been added with a specific example.

Reviewer 2 Report

Comments and Suggestions for Authors

In this manuscript, the authors develop and characterize the prototype of a greenhouse in which light irradiation from an external source is controlled, in real time, by tuning the transparency of the PDLC walls. PDLC (polymer-dispersed liquid crystals) are thin films of thermotropic liquid crystal droplets suspended in a solid polymer matrix. Normally, light passing through is scattered, leading to translucent behavior and to a limited light transmittance. Application of an electric field across the film leads to the orientation of the liquid crystal molecules and to a radical change in the optical properties, with the film increasing its light transmittance and, eventually, being nearly transparent.

The authors use a commercial PDLC film for their project, demonstrating using standard control models that the wall transmittance can be controlled using a feedback loop that keeps a target illumination of the internal space.

The text is clearly written, the experiments are clearly performed, and discussion and conclusions are well supported by the data provided. This work will be of interest to readers wanting to implement applications based on the control of PDLC devices, whose performance as a window with controllable transparency has been demonstrated.

Despite my favorable comments, there are some minor aspects that need to be addressed to improve the quality of the manuscript, as detailed below:

1. In the abstract, the acronym “PI” is used before being defined. Even though it is included in the acronym list at the end of the manuscript, it should be defined the first time it is used.
2. In Eq. 1, the different terms should be defined (parameters, variables, etc)
3. Through the manuscript, the values of many empirically determined parameters are given, but without mentioning their precision. All parameters should be given with an estimation of their error and, correspondingly, with the correct number of significant digits. This applies to Tables 1 through 4.
4. I am confused about the units of the light disturbance, “d”. I see the unit “micromole”, which seems surprising to me. Please revise and correct/comment to clarify.
5. In line 206, I do not understand the meaning of “ …there is also a ZERO-ORDER approximately 0.1 V jump …”. Please, clarify.
6. In Fig. 10, please define what the dashed and solid lines are.
7. I am surprised about the large number of fitting parameters in eq3 and eq4, considering that the data to be fitter has a very smooth and featureless profile. It is crucial to provide the uncertainties of the fit parameter. I am convinced that some of them will have huge uncertainties, indicating that the model has too many parameters that cannot be differentiated. Please, revise, correct, and discuss.

Author Response

The authors thank the reviewer for the insightful comments.  We endeavored to address all the comments in the revised draft and in the discussion below:

1. In the abstract, the acronym “PI” is used before being defined. Even though it is included in the acronym list at the end of the manuscript, it should be defined the first time it is used.

Thank you for the insightful comment, it is corrected.

1. In Eq. 1, the different terms should be defined (parameters, variables, etc)

Thank you for catching the oversighted, it has been corrected.

1. Through the manuscript, the values of many empirically determined parameters are given, but without mentioning their precision. All parameters should be given with an estimation of their error and, correspondingly, with the correct number of significant digits. This applies to Tables 1 through 4.

Table 1 are chosen values. Implemented for Table 2 as it is measured values. Table 3 and 4 are curve fits, digits seen are the chosen significant figures of the values.

1. I am confused about the units of the light disturbance, “d”. I see the unit “micromole”, which seems surprising to me. Please revise and correct/comment to clarify.

Thank you for your comment. Further clarification was added in the paper.

1. In line 206, I do not understand the meaning of “ …there is also a ZERO-ORDER approximately 0.1 V jump …”. Please, clarify.

Thank you for the insightful comment. The section was rewritten to make it more clear.

1. In Fig. 10, please define what the dashed and solid lines are.

Thank you for pointing this out. A definition was added.

1. I am surprised about the large number of fitting parameters in eq3 and eq4, considering that the data to be fitter has a very smooth and featureless profile. It is crucial to provide the uncertainties of the fit parameter. I am convinced that some of them will have huge uncertainties, indicating that the model has too many parameters that cannot be differentiated. Please, revise, correct, and discuss.

Thank you for the insightful comment. The coefficient of determination was added for equation3. However equation4 is a third order transfer function with static gain of unity which was needed to mimic the dynamic behavior of the PDLC.

Reviewer 3 Report

Comments and Suggestions for Authors

A Control Oriented Model for PDLC Films as an Actuator for Natural Light Control
Alexander H. Pesh and Chiara Vetter

The paper describes the design, simulation and experimental setup of control system of natural light using a PI controller. The paper is very well organized and written.

Please, consider the following comments/suggestions. 

1) Abstract, last line: ''This shows that PDLCs can be effective for the control of light''. Perhaps what is meant here is that the proposed control framework allows for continuous control of light. It is already known that PDLCs can effective for the purpose of the control of light, although with on-off control as the authors refer along lines 11-12.

2) Line 65: it is pointed out that there are few works on continuous control of light, which is needed for optimized light level for plant growth. Plant growth dynamics are slower in comparison with light changes. Please, be more specific on why there is need for continuous control, and a simple on-off control scheme is not sufficient.

3) Line 97: ''... minimizes the error and optimizes the PDLC performance for maximal plant growth.'' Clarify the idea of optimizing the PDLC performance. Is this related to the fact that when using a PI control loop, and not a simple on-off controller, and the setpoint of the PI is as well optimized, let us say in function of a pre-determined light profile for the different growth stages?

4) Figure 11 shows a nonlinear variation of the transmitance T with respect to the PWM signal. The Photosynthetically Active Radiation (PAR) measured under the PDLC, denoted by L, is given by the product between T and PAR measured with no PDLC (see lines 242-243). It is expected that the relation between L and PWM is also nonlinear. As for the normalized values of L in Figure 12, were these obtained for the range of values of interest for this system? Also, although the nonlinear relationship mentioned above, it is assumed the description of the dynamics is good enough. Please, clarify this.

5) Line 280-281:  the numerical integrator ode45 is mentioned here. Does this mean a state-space realization with 3 state variables was obtained from the linear model (4)?

6) Lines 320-321: ''The novel model is used for numerical simulations of the closed-loop; ...'' Clarify the role of the model in the development and design of the proposed control framework. On the title one reads ''A Control Oriented Model...'' The title conveys the idea of a model based control strategy. Obviously a model is very useful to simulate a wider range of scenarios and can be used as a decision support tool in the process of designing the control strategy. This should be highlighted in the paper. Another question, could this model be used online in a control model based structure? 

Author Response

The authors would like to thank the reviewer for the insightful comments.  We have endeavored to address all of them in the revised draft and in the comments below:

1) Abstract, last line: ''This shows that PDLCs can be effective for the control of light''. Perhaps what is meant here is that the proposed control framework allows for continuous control of light. It is already known that PDLCs can effective for the purpose of the control of light, although with on-off control as the authors refer along lines 11-12.

Thank you for the insightful comment. An adjustment in the text was made.

2) Line 65: it is pointed out that there are few works on continuous control of light, which is needed for optimized light level for plant growth. Plant growth dynamics are slower in comparison with light changes. Please, be more specific on why there is need for continuous control, and a simple on-off control scheme is not sufficient.

Thank you for the insightful comment. Further clarification was added in the paper.

3) Line 97: ''... minimizes the error and optimizes the PDLC performance for maximal plant growth.'' Clarify the idea of optimizing the PDLC performance. Is this related to the fact that when using a PI control loop, and not a simple on-off controller, and the setpoint of the PI is as well optimized, let us say in function of a pre-determined light profile for the different growth stages?

Thank you for pointing this out. Further clarification was added in the paper.

4) Figure 11 shows a nonlinear variation of the transmittance T with respect to the PWM signal. The Photosynthetically Active Radiation (PAR) measured under the PDLC, denoted by L, is given by the product between T and PAR measured with no PDLC (see lines 242-243). It is expected that the relation between L and PWM is also nonlinear. As for the normalized values of L in Figure 12, were these obtained for the range of values of interest for this system? Also, although the nonlinear relationship mentioned above, it is assumed the description of the dynamics is good enough. Please, clarify this.

Yes, the selected range is characteristic of average values of the actuation.

5) Line 280-281:  the numerical integrator ode45 is mentioned here. Does this mean a state-space realization with 3 state variables was obtained from the linear model (4)?

Yes, that is correctly stated.

6) Lines 320-321: ''The novel model is used for numerical simulations of the closed-loop; ...'' Clarify the role of the model in the development and design of the proposed control framework. On the title one reads ''A Control Oriented Model...'' The title conveys the idea of a model based control strategy. Obviously a model is very useful to simulate a wider range of scenarios and can be used as a decision support tool in the process of designing the control strategy. This should be highlighted in the paper. Another question, could this model be used online in a control model based structure? 

Thank you for the insightful comment. The last section was increased by adding further discussion how the simulation can be useful for future work.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

I appreciate the author's response and modification for the previous comments.

I can suggest the publication of this manuscript.

Again, thanks for the author's effort.

Author Response

The authors would like to thank the reviewer for the feedback which helped making a stronger paper.