rapidMQI

Start here

1. Load a photograph of the wall, or take one. Stand square to the wall if you can: a face seen at a sharp angle cannot be measured.
2. Set the scale. Pick Scale, click the two ends of something whose length you know — a brick course, a door opening, a folding rule left against the wall — and type that length. Everything measured afterwards is in metres.
3. Work through the seven parameters. Each one shows the criteria of the paper, and you choose Fulfilled, Partially fulfilled or Not fulfilled. For the leaf connection and the vertical joints you can trace the path that decides it; for the block dimensions you can measure a handful of blocks.
4. Say what the inside is like. On the right are two more views: a section through the thickness and a typical block. Photograph them if anything is exposed; if nothing is, pick the drawing that matches what the wall is understood to be.
5. Read the result. Three indices, three categories, and the mechanical properties the correlation curves give.
6. Save the survey to a file, and print the report.

A survey of one wall takes a few minutes once the photograph is scaled. A building is several walls: add a panel for each, and the Project pane compares them.

The index

The Masonry Quality Index asks a single question — was this wall built according to the rule of the art? — and breaks it into seven observable properties. Each is judged Fulfilled (F), Partially fulfilled (PF) or Not fulfilled (NF), each outcome carries a number from Table 8 of the paper, and six of the seven are added together:

The seventh, SM, is not added: it multiplies. The state of the blocks themselves scales the quality of everything that was done with them, so a wall of decayed stone cannot score highly however carefully it was coursed. SM runs from 0.3 to 1, the sum from 0 to 10, and the index with it.

The numbers differ for the three loading conditions, because the same wall is not equally good at three different jobs. The connection between the leaves is worth 1 point out of 10 when the wall is merely carrying weight and 3 when it is being pushed out of plane; the staggering of the vertical joints is worth 1 vertically and 2 in plane. So the index is computed three times, and a wall can be class A for vertical actions and class C in plane.

The seven parameters

The columns are what each parameter is worth when it is fulfilled, under vertical, in-plane and out-of-plane actions. They sum to 10 in each case. The application shows the same three numbers beside every choice, so it is always visible what an answer is worth.

Reading a photograph

The photograph is not decoration. Three of the seven parameters can be measured on it rather than merely judged, and the marks stay on the picture so that a reader of the report can see what the outcome was based on.

Scale

One known distance turns pixels into metres. A chequered bar one metre long then appears on the photograph, and every measurement is reported in centimetres or metres.

Ruler, for SD

Click the two ends of a block. Table 2 asks about more than half of the elements, so what it wants is the median of a representative sample: measure a handful and the application reports the median and which band it falls in.

Mark

A point on the photograph, attached to the parameter being assessed — the breach where the section is visible, the one course of bricks that makes the bed joints continuous, the decayed stones that settle SM.

Three views of the same wall

A photograph of a wall face answers four of the seven parameters and cannot answer two of the others. The connection between the leaves is inside the wall; the shape of a block is a three-dimensional fact that a flat view states poorly. So a panel in rapidMQI holds three views, and the strip above the photograph switches between them:

Each view carries its own scale and its own marks. A section is photographed from half a metre and a face from ten, so one scale for both would make every measurement on one of them wrong. Set the scale once per photograph and the tools follow.

When there is nothing to photograph

Most walls expose no section at all, and that is the ordinary case rather than the awkward one. For the section and for the block, the application offers three drawings, one for each outcome: two unconnected leaves with a straight joint down the middle, two leaves with some headers, or stones reaching from face to face; a pebble, a roughly squared lump, or a cut ashlar. Picking the one the wall is understood to be assesses the parameter and keeps the drawing.

Dimensions, written under the drawing

Under each of the two drawings is a row of fields, and they are the geometry row of the paper's own data sheets: the three sides of a typical block — l the length, h the height, s the depth into the wall — and, for the section, the thickness t of the wall and the number of headers per square metre. Write them the way a surveyor writes them, as a range where a range is the truth: 24÷32 is as good an answer as 30. They are quoted on the drawing as they are typed, and they are printed on the data sheet.

Quotes on a photograph

When there is something to photograph, the same dimensions can be measured on it. The Ruler draws a dimension line with its value on it, and a quote can be given a name, so that a photograph of a section carries its thickness the way a drawing would: t = 118 cm. A quote drawn on a section belongs to the leaf connection; a block measured on the face is a sample for the dimensions of the elements. The report prints the photographs with those quotes on them.

The minimum length M_l

Two of the parameters have a quantitative criterion, and it is the same idea in both cases. Take two points about a metre apart and find the shortest path that joins them while running only through mortar joints, never through a block. The ratio of that path to the straight distance is M_l:

A wall of large, well interlocked blocks forces the path to wander, and M_l is large. A wall whose joints line up lets it run almost straight, and M_l approaches 1 — which is another way of saying that a crack would run just as easily.

Pick the Path tool and click along the joints from one point to the other, then Finish. The application divides by the straight distance, says which outcome that implies, and warns if the two ends are closer than the half metre below which the ratio stops describing the texture of the wall and starts describing one joint.

What cannot be seen

Some parameters cannot be assessed from outside. If no section is exposed, the connection between the leaves is a guess; if nothing has been opened up, so is the mortar behind the pointing. The paper assumes all seven are assigned. rapidMQI allows one to be left as not determinable, and then reports the index as the interval it is known to lie in — the value it would take if the missing parameter turned out to be NF, and the value it would take if it turned out to be F.

The category may then be an interval too, and the application says so rather than choosing. It also says which missing observation would close the widest part of the interval, which is the useful question when there is time to go back to the building once.

The result

Three indices, and a category for each. Category A is a masonry of good behaviour, B of average quality, C inadequate. The thresholds differ by loading condition: out-of-plane actions are judged most strictly, because the out-of-plane mechanism is the one that kills.

The Results pane also takes the index apart: what each parameter contributed, under each loading condition, at the outcome that was chosen. It is worth reading before the number is quoted anywhere.

Mechanical properties

Section 3 of the paper correlates the index with the mechanical parameters of the Italian code table, by building 36 virtual wall panels from that table and its multiplication factors and fitting an exponential through the lower and the upper bound of each range. The result is not a value but a band:

The application draws each band and marks where this wall falls on it, and compares the estimate with whichever row of the code table the wall is declared to be. The comparison asks whether the two bands overlap, not whether they agree: where they do not, either the declared typology or one of the seven outcomes is worth revisiting. The shear modulus G is not correlated in the paper and is derived as E/3, the ratio the code table holds; it is labelled as derived wherever it appears.

Files and reports

A survey is one JSON file, with the photographs inside it, so that it opens complete on any machine six months later. A working copy is also kept in the browser, so that a phone going to sleep in a courtyard does not cost a morning's work — but that copy is a convenience, not the record: save the file.

The report prints two sheets per wall. The first is the data sheet of the paper's own worked examples: the three views, what the wall is made of, the seven outcomes in a strip, and the analysis. The photographs on it carry the marks that were made on them — the traced path with its M_l, the measured blocks, the scale reference — so that a reader can see what an outcome was based on instead of taking it on trust.

The second sheet is what the application shows on the screen: the three indices with their category bars, the table that takes the index apart, the mechanical properties on their correlation curves, the comparison with the code table, the reference values it was checked against and the bands of Table 9. It is the part of a survey anyone else will want to argue with, so it is the part worth printing in full.

Two editorial decisions

Anyone comparing this application against the paper will meet two places where it does not do what the paper appears to say, and both are deliberate.

The categories of Table 9 are the right way round here

Table 9 as printed carries its column headers in the order A, B, C above ranges that run from the lowest index to the highest. Read literally, MQI = 8.5 would be a category C masonry — which contradicts the text, where A is "good behavior of masonry", and contradicts the paper's own worked example in Fig. 12, where 8.5 / 9.0 / 9.5 is classified A / A / A. rapidMQI puts the labels back the right way round: C is the lowest band, A the highest.

The shear strength is read at the in-plane index

The abscissa of Fig. 10 is labelled "MQI value (vertical actions)", but all three data sheets in the paper read τ₀ at the in-plane index instead: Fig. 12 prints 0.133 to 0.182 MPa, which is the curve at 9.0 and not at 8.5, and Figs. 13 and 14 do the same. Three examples agreeing is a rule, and it is the physically sensible one, since shear strength is what governs the in-plane response. rapidMQI follows the data sheets, and the Project pane has a switch for anyone who would rather follow the axis label.

How to cite

The method is not ours. Cite the paper:

The software itself has a CITATION.cff in its repository. It is free software under the MIT licence, and its source is what runs: plain modules, no build step, nothing minified.