Drying kinetics and microclimate data of six foods in a tunnel-type solar dryer

Drying kinetics and microclimate data of six foods in a tunnel-type solar dryer Mass loss and drying-air data recorded in a tunnel-type solar dryer for seven foods dried between July and October 2024. Each record contains the mass of the product on one tray together with the conditions of the drying air at that instant (temperature and relative humidity along the tunnel, outdoor temperature and relative humidity, solar irradiance and the fan setting), so the data can be used to fit or test drying-kinetics models under real, non-isothermal conditions. The dataset contains the raw logged variables together with the derived quantities normally used for modeling: the mass ratio and the dry-basis moisture ratio. Contents Totals: 7 foods, 44 series, 4574 records. A series is one tray of one drying day, which is the experimental unit: the load cell under each tray integrates the mass of the product on that tray only, while the air conditions are shared by the whole chamber. Data format Column Units Description series_id Series identifier: <food>_<YYYYMMDD>_tray<n> food, food_label Machine-readable and human-readable product name date Drying day (YYYY-MM-DD) tray Tray number, 1 to 3, counted from the fan outlet elapsed_min, elapsed_h min, h Time since the first record of the series, from the acquisition counter. Records are taken every 5 min server_timestamp Date and time as stored by the local server. Approximate: see the note below mass_g g Mass of the product on the tray initial_mass_g g Mass of the product on that tray at the start of the series mass_ratio mass_g / initial_mass_g moisture_content_dry_basis kg water / kg dry matter $X = \frac{M - m_d}{m_d}$, with $m_d = M_0 (1 - w_0)$ moisture_ratio_dry_basis $MR = \frac{X - X_e}{X_0 - X_e}$. See Moisture contents below mass_increase_flag 0/1 1 if the mass is larger than in the previous record, which is not physically possible during drying mr_out_of_range_flag 0/1 1 if moisture_ratio_dry_basis falls outside [0, 1] T_tray_C, RH_tray_pct °C, % Air temperature and relative humidity next to the tray: the mean of the two probes that flank it (probes 1–2 for tray 1, 2–3 for tray 2, 3–4 for tray 3) T_ambient_C, RH_ambient_pct °C, % Outdoor air temperature and relative humidity irradiance_W_m2 W m⁻² Global solar irradiance pwm Duty-cycle setting commanded to the fan, as logged T_probe1_C to T_probe4_C °C The four temperature probes along the tunnel, in order from the fan outlet RH_probe1_pct to RH_probe4_pct % The four relative-humidity probes, same positions One row per tray and per sampling instant. All files share the same columns. The four probe columns are repeated on the three tray rows of the same instant, because the probes are shared by the chamber. `T_tray_C` and `RH_tray_pct` are the only derived air variables; every other air column is as logged. Note on the time columns The acquisition system (an ESP32) sampled every 5 minutes and sent the records to a local server, which is the one that wrote the date and time. Because the server assigned those timestamps once the transfer was finished, the difference between two consecutive `server_timestamp` values does not reproduce the sampling interval, and over a whole run it accumulates a stretch of about 15 %. **Use `elapsed_min` or `elapsed_h` as the time variable**, which come from the counter of the acquisition system and are exactly 5 minutes apart. `server_timestamp` is kept only as an approximate reference for the time of day, for example, to relate a record to the position of the sun. Note on the multi-day batches Apple, banana, beef, lemon and orange peel were dried in a single day. The two grape batches took several days, and the product stayed in the dryer overnight, so their records include night hours (irradiance near zero or slightly negative, see below). For these batches `initial_mass_g` is always the fresh mass at the start of the **batch**,not the mass at the start of each day, so `mass_ratio` is referred to the fresh product throughout. What restarts every day is `elapsed_min`, because a series is one tray-day.The dataset does not provide a reliable continuous time axis across days: the counter of the acquisition system pauses when logging stops and the server clock is only approximate, so the elapsed time between the end of one day and the start of the next has to be reconstructed from `date` and, approximately, from `server_timestamp`. Experimental setup Tunnel-type solar dryer with a square drying chamber covered with translucent glass, three racks, trays of 0.3 m × 0.3 m, a water–air heat exchanger of 0.16 m² coupled to a 12 V axial fan, and four solar collectors of 2.5 m² each heating the water. Air velocity through the chamber 1.5 m s⁻¹. Instruments: Instrument Model Relevant specification Temperature / relative humidity Sensirion SHT4x, four probes 40 cm apart ±0.2 °C, ±2 %RH (typical) Mass HT Sensor Technology TAL220, 5 kg parallel-beam strain gauges, one per tray Combined error ±0.05 %FS (±2.5 g), non-linearity ±0.05 %FS, repeatability ±0.03 %FS. Calibrated with a certified 300 g mass and tared before each test Global solar irradiance Kipp & Zonen CMP6 pyranometer 285–2800 nm, response time 12 s Products were bought at a local market, cleaned, disinfected, and sliced when necessary, then distributed on the trays; drying ran during daylight hours, one batch per day. Moisture contents moisture_content_dry_basis` and `moisture_ratio_dry_basis` are derived from the mass records with the initial and final moisture contents reported for these same products in Table 4 of Rodriguez-Ortiz et al. (2026), on a wet basis: Food Initial moisture (%) Moisture of the dried product (%) Apple 81.38 13.10 Banana 76.48 10.76 Grape 52.06 10.28 Beef 82.10 6.91 Orange peel 84.27 1.61 Lemon 68.48 7.22 `Xe` is taken as the moisture content of the fully dried product, so `MR` is an affine function of `mass_ratio`. For beef this gives `MR = (mass_ratio − 0.192)/0.808`. Since the transformation is affine, any model ranking obtained with `mass_ratio` also holds for `MR`; only the scale of absolute errors changes. Users who prefer their own moisture determination can recompute both columns from `mass_ratio` and their own `w0`. Known issues These are stated explicitly so that anyone reusing the data can decide how to handle them. Load-cell noise, mainly on tray 3. The mass signal is not monotonically decreasing. In the beef series, the fraction of records that show an increase in mass is 23–38 % on tray 3, against 2–18 % on tray 1 and 6–20 % on tray 2. Tray 3 is the farthest from the fan outlet and its cell is the noisiest of the three. Use `mass_increase_flag` to locate these records, and `series_index.csv` for the fraction per series. Server_timestamp` is not a reliable clock. See Note on the time columns above. The time variable to use is `elapsed_min`. Grape moisture content. With the tabulated initial moisture of 52.06%, the observed mass losses for the grape series imply a moisture content below the dry-matter content, yielding negative `moisture_ratio_dry_basis` values (down to −0.28). Either the tabulated value or the final moisture of those series is inconsistent, so for grape the `mass_ratio` column is the reliable quantity. Records affected are flagged with `mr_out_of_range_flag`. Values slightly above 1. A few records at the start of a series exceed `mass_ratio = 1` because of load-cell drift and taring, which puts `moisture_ratio_dry_basis` marginally above 1 (up to 1.12 in beef). They are flagged and have not been clipped.